# The Robots Are Coming for Your Job, and That's Okay

Elmer Thomas, Twilio · Maria Bermudez

AI Engineer World's Fair 2025

[Talk page](https://ai.engineer/talks/pSqpC7fFLZA-robots-are-coming-your-job-thats-okay) · [Video](https://www.youtube.com/watch?v=pSqpC7fFLZA)

## The Robots Are Coming for Your Job, and That's Okay

Twilio’s documentation workflow uses focused AI helpers to propose edits, generate metadata and simplify prose, while validation and human review keep publication under team control.

Familiarity with Markdown, pull requests and automated checks will help you follow the documentation workflow.

### A small docs team facing a flood of requests

[0:11](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=11s)

A small documentation team sits downstream of a much larger organization: Jira tickets keep arriving, first drafts need corrections, and routine grooming competes with work that requires editorial judgment. The goal for Twilio’s team is to use AI to absorb some of that repetitive work without replacing the people responsible for clarity and accuracy.

Elmer Thomas, Principal Developer Educator at Twilio, introduces the workflow alongside Maria Bermudez, a lead developer of the team’s AI Docs Buddy. Thomas describes error-prone first drafts arriving from more than 100 product teams. Style checks, image alt text and SEO metadata consume additional time. Generative AI offers help with those chores, but unconstrained generation introduces another problem: plausible content that may be wrong.

![Slide titled “The Pain” shows a docs team behind a help desk surrounded by people holding Jira papers.](https://ai.engineer/api/talk-essay-media/4723b963575cfe706d35cf25781d0e99609254fdc43595db79568ca2179c4413)

The docs team faces a crowd of Jira requests.

### Give each helper a narrow job

[1:25](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=85s)

The team organizes the work into six single-purpose helpers behind a simple Next.js front end. Each has a bounded output that a person can inspect.

| Helper | Assigned job |
| --- | --- |
| Automated editor | Grammar, formatting and accuracy edits |
| Image alt text generator | Accessibility descriptions |
| Jargon simplifier | Developer language into plain English |
| SEO metadata generator | Titles and descriptions within character limits |
| Docs outline builder | Navigation and structure recommendations; coming soon |
| Slack bot | Help-channel request triage |

The outline builder is planned work at the time of the recording. The broader division of labor is already clear: repetitive jobs go to specialized helpers, leaving humans to concentrate on judgment and clarity.

### Turn generated changes into reviewable proposals

[2:16](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=136s)

**Repeatable, high-volume work with little need for creativity is the starting point.** A request enters through the Next.js UI and reaches a custom agent, with the model chosen for the particular job. The agent uses the team’s style guide and rubric, retrieved from Airtable so collaborators can maintain the guidance together. Editorial policy therefore has an explicit home outside the individual request.

The proposed changes then pass through a validation layer containing [Vale](https://github.com/vale-cli/vale) linting and CI/CD tests. Vale checks configurable prose rules; it does not establish whether a product claim is true. A GitHub pull request makes the changes available for scrutiny through [code-owner review](https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-code-owners). GitHub treats requesting that review and requiring approval before merge as separate settings.

Product and engineering reviewers usually participate as well, and a human merges only after accepting the changes. **Generating an edit does not authorize publication.** Thomas initially describes the layers as crushing hallucinations, then corrects himself: the claim is that they significantly reduce them. The protection comes from combining automated checks with several people’s judgment, rather than asking any single layer to guarantee correctness.

![“Under the Hood” diagram links Next.js UI to a custom GPT-4/o1 agent, validation layer, GitHub PR, and human in the loop.](https://ai.engineer/api/talk-essay-media/6c55c673afd007e64a7ad19971e455486d14a9fafdbee372f0b01174f079c849)

The request flow connects the Next.js UI, agent, validation, GitHub PR, and human review.

### Inspect both the edit and its reason

[3:51](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=231s)

Bermudez begins in the application’s overview and release notes page, which the team uses Copilot to generate. From there, she opens the automated editor. It accepts an MDX file, a Markdown file or a live URL; she chooses the URL option. This lets the editing pass start with an existing published page rather than requiring someone to prepare a separate text input.

The editor uses **o1**, and the interface displays a loading graphic while it works. Bermudez reports that experimentation identified it as the best model for consistently applying this team’s style guide and rubric. That choice carries a visible latency cost in the interaction.

The result offers two complementary ways to review the proposal. A diff shows what changed. The Changes Made tab provides the original text, revised text and an explanation connecting the change to a style-guide instruction or rubric item. The reviewer can inspect both the textual effect and the rule offered as its justification before deciding whether to accept it.

### Run a separate metadata pass

[4:53](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=293s)

The editor’s output also exposes a limit: it sometimes catches a missing SEO description and sometimes does not. Bermudez moves the same page into the dedicated SEO metadata generator. Specialization gives the missing field its own explicit task instead of relying on the general editing pass to notice every omission.

She uploads a file and returns to the Conversational Intelligence page that was just edited. The generator produces meta titles by default—users must specify if they do not want them—and a meta description that accounts for character limits. The handoff is performed by Bermudez through the interface, keeping the metadata operation separate and visible.

### Generate alt text across multiple pages

[5:52](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=352s)

The alt text generator also supports a live URL, but Bermudez demonstrates selecting multiple pages and generating descriptions for all their images in one action. Batch input reduces the repeated work of submitting each image separately.

The results display image filenames and previews beside generated text. Bermudez calls attention to the quick completion and to a second practical requirement: the output conforms to the documentation platform’s expected format. Producing a description is only part of the task; producing one in the form the publishing workflow accepts makes the result easier to use.

![Image Alt Text Generator results show image filenames, previews, and generated text in adjacent panels.](https://ai.engineer/api/talk-essay-media/9ecd57c9c83ca3358500d2b35f63ec3a4fb797a90983d55906149d9738fb3be7)

Generated alt text appears beside uploaded image previews.

### Make revisions easy to reuse in a pull request

[6:25](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=385s)

The jargon simplifier serves both authors drafting a page and reviewers working through a pull request. Bermudez pastes prepared text and selects Simplify. The result includes a diff, like the automated editor, plus a separate revised-text tab. A reviewer can copy that text into a pull request comment or use it to edit the file directly.

![Jargon Simplifier displays original text above the selected “Revised Text & Explanation” tab, with revised text, a copy icon, and a collapsed “Explained Jargon” section.](https://ai.engineer/api/talk-essay-media/91a2f9566fb08cc9d18c13714e2be514d5e8e9fde965feba7ca1d9972ba31f61)

The jargon simplifier presents revised text with a copy control.

For example, a Markdown review suggestion can make a proposed simplification directly usable:

```markdown
Replace:
Utilize the configuration interface to modify notification preferences.

With:
Use settings to change your notification preferences.
```

The useful handoff is a concrete revision that a reviewer can accept, adjust or reject. It remains a proposal until someone applies it.

These helpers are not yet an autonomous chain. Bermudez closes the walkthrough by saying the team wants the agents to communicate with one another, but substantial work remains. The demonstrated workflow consists of focused tools with a person moving between them.

### Maintain quality through distinct feedback loops

[7:20](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=440s)

Tools alone do not guarantee quality. Thomas separates the remaining risks by the kind of review they need:

- **Hallucinations:** Combine Vale linting and CI tests with scrutiny from human stakeholders.
- **Bias:** Use dataset tests and prompt audits.
- **Stakeholder misalignment:** Review pull requests weekly, accelerating the cycle to days or hours when needed.

These checks cover different failure modes; passing a style check does not settle a factual question or establish that stakeholders agree with the result.

Slack feedback loops with product managers and engineering provide another route for corrections. The team uses those exchanges to tune prompts continuously. Quality maintenance is therefore an ongoing collaboration around actual outputs, rather than a one-time prompt-writing exercise.

### Start with one throughput bottleneck

[8:12](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=492s)

The adoption process begins smaller than the full set of helpers:

1. Identify one pain point that is limiting throughput.
2. Choose a single task that is repeatable and rule-based.
3. Loop with its users at least weekly.

Ship, measure and refine that workflow. Thomas expects several such improvements to accumulate into greater team velocity; the recording offers that as the direction of travel, without a measured team-wide gain. The practical starting point is one bounded job whose outputs people can review and whose shortcomings can feed the next iteration.

## Resources

- [Docs AI Buddy Prompt Library](https://github.com/twilio-labs/docs-ai-buddy) — Reusable prompts for documentation editing, jargon simplification, alt text and SEO metadata, with examples and integration guidance.
- [Twilio's Docs AI Buddy prompt release](https://www.twilio.com/en-us/blog/developers/open-source-docs-ai-buddy-prompts) — Twilio's announcement explaining how to adopt and contribute to its public documentation prompt templates.
- [Making Alt Text Fast](https://www.twilio.com/en-us/blog/developers/docs-accessibility-automation-llm-prompts) — Maria Bermudez explains Twilio's batch alt-text workflow, human review and subsequently reported production results.
- [Vale prose linter](https://github.com/vale-cli/vale) — A configurable prose linter for enforcing editorial rules while respecting document markup.
- [GitHub code-owner reviews](https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-code-owners) — How to assign responsible reviewers and require code-owner approval before merging.

## Slides

### The Robots Are Coming for Your Job, and That’s Okay

Elmer Thomas introduces how Twilio’s small docs team uses AI agents to improve workflows. Maria Bermudez will lead a live agent walkthrough.

[0:00](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=0s) [Twilio title slide naming presenters Elmer Thomas and Maria Bermudez.](https://assets.ai.engineer/sites/aiecode2025/sha256/f4f93b468f03b36ff1a04f2504a02d2436789dd70e1945435596fb64398341da.webp)

twilio

The Robots Are Coming for Your Job, and That’s Okay.

Elmer Thomas
Developer Educator at Twilio

Maria Bermudez
Developer Educator at Twilio

2025 TWILIO INC. ALL RIGHTS RESERVED

### The Pain

A small docs team faces a flood of Jira tickets, error-prone drafts from 100-plus product teams, time spent on style, alt text and SEO, and the risk of GenAI hallucinations.

[1:04](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=64s) [A crowded help desk surrounds a documentation team with Jira requests.](https://assets.ai.engineer/sites/aiecode2025/sha256/72647db226de3b02dcebb75431f895ef8259c9f44f65ef6c1c07d43fa5fe9e2d.webp)

The Pain

Signs: JIRA (repeated)
DOCS TEAM
HELP DESK

2025 TWILIO INC. ALL RIGHTS RESERVED

### Meet the Twilio AI Docs Buddy

Twilio splits documentation work across six single-purpose agents, from editing to Slack triage. The outline builder is coming soon; Slack BatBot is in internal beta.

[1:27](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=87s) [Six AI documentation tools cover editing, alt text, jargon, SEO metadata, outlines, and Slack help requests.](https://assets.ai.engineer/sites/aiecode2025/sha256/3d7c1de5babd291340322572826ea36f69699a9eb796e7ba6900f2cf679762c8.webp)

Meet the Twilio AI Docs Buddy

Automated Editor
Automatically edits and improves documentation according to DevEd's style guide and quality rubric. This agent can fix grammar, spelling, formatting, and ensure technical accuracy.

Image Alt-Text
Analyzes images and generates accessible alt text descriptions in accordance with DevEd standards.

Jargon Simplifier
This agent checks for technical jargon, simplifies complex terms, and flags or explains them as needed.

SEO Metadata
This agent analyzes the content of a specific page and creates an SEO metadata title and description, optimized for search engines, that follow DevEd standards.

Docs Outline Builder
Recommends documentation outline and navigation structure. Coming soon.

Slack BatBot
Monitors our Slack help channel to triage requests and assist with onboarding to our docs-as-code platform. Internal Beta.

2025 TWILIO INC. ALL RIGHTS RESERVED

### Under the Hood

A Next.js UI sends requests to a custom agent, then a validation layer and GitHub PR. Vale linting, CI/CD tests, and human review check suggested changes before a person merges them.

[2:34](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=154s) [A workflow connects a Next.js interface to a custom agent, validation, a GitHub pull request, and human review.](https://assets.ai.engineer/sites/aiecode2025/sha256/537d2205eea57f01375215da66de5ebe47aa42cab73f356110dfd86341f5b706.webp)

Under the Hood

NEXT.JS UI → CUSTOM GPT-4/o1 AGENT → VALIDATION LAYER → GITHUB PR → HUMAN IN THE LOOP

2025 TWILIO INC. ALL RIGHTS RESERVED

### Docs AI Buddy overview

Maria says Copilot generated the overview and release notes page. It links to documentation agents, with the Docs Outline Builder marked as coming soon.

[3:50](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=230s) [The Docs AI Buddy overview lists documentation agents and links to their tools.](https://assets.ai.engineer/sites/aiecode2025/sha256/81c3cb948a09aad57a68cd949bdd2ea17a0d90fc5f25c21789873dea1d7f9c88.webp)

Twilio Docs AI Buddy

Overview
Twilio Docs AI Buddy provides AI-powered assistance for creating and maintaining high-quality technical documentation. Our suite of specialized agents helps you improve document quality, ensure compliance with the DevEd style guide and rubric, and streamline documentation workflows.

Available AI Agents

Automated Document Editor
Automatically edits and improves documentation according to DevEd's style guide and quality rubric. This agent can fix grammar, spelling, formatting, and ensure technical accuracy.
Use automated editor →

Image Alt Text Generator
Analyzes images and generates accessible alt text descriptions in accordance with DevEd standards.
Use alt text generator →

Jargon Simplifier
This agent checks for technical jargon, simplifies complex terms, and flags or explains them as needed.
Use jargon simplifier →

SEO Metadata Generator
This agent analyzes the content of a specific page and creates an SEO metadata title and description, optimized for search engines, that follow DevEd sta[illegible]
Use SEO metadata generator →

Docs Outline Builder
Recommends documentation outline and navigation structure. Coming soon.

Release Notes

Interface text:
docs-ai-buddy.vercel.app/agents
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

### Auto-edited document comparison

The editor highlights deletions in red and additions in green. The Changes Made tab lists explanations alongside the relevant style guidance and rubric items.

[4:28](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=268s) [Original and edited document text displayed side by side, with deletions in red and additions in green.](https://assets.ai.engineer/sites/aiecode2025/sha256/9b8ff78d521e6b845d1d12cc7ac6dbb48fc599a0ac7920757b5e83dbf7b928fb.webp)

Auto-Edited Document

Auto-edit another document
Create a pull request
Comparison
Changes Made
Copy Edited Content

Original
Expand 6 lines...
7  index_by_search_engines: false
8  ---
9
10 - ## General Limitations and Caveats
11
12   - Conversational Intelligence is not PCI or HIPAA compliant.
13 - - Encrypted recordings and recordings of PCI[illegible]
       enabled accounts are not supported.
14 - - Twilio Recordings
        with external storage are only accepted using the `MediaUrl`
       on the Transcript creation.
15
16   ## Transcripts App

Edited
7  index_by_search_engines: false
8  ---
9
10 + ## General limitations and caveats
11
12   - Conversational Intelligence is not PCI or HIPAA compliant.
13 + - Encrypted recordings and recordings [illegible]
       enabled accounts are not supported.
14 + - Twilio recordings
        with external storage are only acc[illegible]
       parameter during Transcript creation.
15
16   ## Transcripts App

Interface text:
docs-ai-buddy.vercel.app/agents/automated-editor
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

### Expanded document comparison

The automated editor shows the original and edited document side by side, highlighting deletions in red and additions in green. The Changes Made tab lists explanations and the relevant style guidance and rubric items.

[4:37](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=277s) [Expanded original and edited document comparison showing metadata and highlighted changes.](https://assets.ai.engineer/sites/aiecode2025/sha256/ab0eec464074759527723a0fed104a4ea84a4cb564133988daaa57cda746d7fa.webp)

Auto-Edited Document

Auto-edit another document
Create a pull request

Comparison
Changes Made
Copy Edited Content

Original
1  ---
2  title: Limitations
3  seo_description: ""
4  show_help_footer: false
5  show_breadcrumbs: false
6  code_column: false
7  index_by_search_engines: false
8  ---
9
10 - ## General Limitations and Caveats
11
12   - Conversational Intelligence is not PCI or HIPAA compliant.
13 - - Encrypted recordings and recordings of PCI
       enabled accounts are not supported.
14 - - Twilio Recordings

Edited
1  ---
2  title: Limitations
3  seo_description: ""
4  show_help_footer: false
5  show_breadcrumbs: false
6  code_column: false
7  index_by_search_engines: false
8  ---
9
10 + ## General limitations and caveats
11
12   - Conversational Intelligence is not [illegible].
13 + - Encrypted recordings and recordings [illegible]
       enabled accounts are not supported.
14 + - Twilio recordings

Interface text:
docs-ai-buddy.vercel.app/agents/automated-editor
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)

### Explanations for editorial changes

The Changes Made tab pairs edits with original and revised text, explanations, and style guidance. The examples cover sentence case and hyphenating “PCI-enabled”; the heading example also cites a rubric item.

[4:42](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=282s) [Changes Made tab lists original and revised headings, explanations, style guidance, and a hyphenation change.](https://assets.ai.engineer/sites/aiecode2025/sha256/e47680c41bf282c969a4e87fe25f777854f1deabb0bec11933ea17963624ac70.webp)

1. Changed headings to sentence case where appropriate.

Original text:
- "## General Limitations and Caveats"
- "## Transcripts App"
- "## Transcription"
- "## Language Operators"
- "## Transcription Viewer"

Revised Text:
- "## General limitations and caveats"
- "## Transcripts App"
- "## Transcription"
- "## Language Operators"
- "## Transcription Viewer"

Explanation: Changed headings to sentence case to comply with style guidelines. Proper nouns remain capitalized.
Style Guidance: "Headings and subheadings should use sentence case unless they are proper nouns."
Rubric: "Are page headings (H2, H3, etc.) used to represent document structure? Are they using sentence-case?"

2. Hyphenated "PCI enabled" to "PCI-enabled".

Original text: "Encrypted recordings and recordings of PCI enabled accounts are not supported."
Revised Text: "Encrypted recordings and recordings of PCI-enabled accounts are not supported."
Explanation: Added a hyphen to form a compound adjective, enhancing clarity.
Style Guidance: "Hyphenate compound adjectives before nouns."

Interface text:
docs-ai-buddy.vercel.app/agents/automated-editor
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Auto-edit another document
Create a pull request
Comparison
Changes Made

### SEO metadata generator

The presenter uploads the page she just edited and prepares to generate SEO metadata. The form also accepts pasted content and optional context, with a description-only setting recommended to avoid URL changes.

[5:10](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=310s) [SEO metadata generator with fields for page content and optional context, plus a description-only option.](https://assets.ai.engineer/sites/aiecode2025/sha256/2d910f86e01d5ce7274fc6296c79dd36e77b6472b8b2ea752d661633f8d87e3c.webp)

Paste Content
Upload File

Paste your page content here...

Additional context (optional)
Add any additional context or specific requirements for the SEO metadata (e.g., 'Focus on JavaScript developers', 'Emphasize security features')

Generate description only (recommended to avoid URL changes)
Generate SEO Metadata

Interface text:
docs-ai-buddy.vercel.app/agents/seo-metadata-generator
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

[5:30](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=330s) [SEO metadata generator with a document uploaded and optional context and description-only controls.](https://assets.ai.engineer/sites/aiecode2025/sha256/1344825449f536be24704844c3d273354b28c428d32133edb6d28b31c5b22799.webp)

Paste Content
Upload File

limitations.mdx uploaded
Browse Files
Supported formats: .txt, .md, .mdx, .html, .json, .js, .jsx, .ts, .tsx

Additional context (optional)
Add any additional context or specific requirements for the SEO metadata (e.g., 'Focus on JavaScript developers', 'Emphasize security features')

Generate description only (recommended to avoid URL changes)
Generate SEO Metadata

Interface text:
docs-ai-buddy.vercel.app/agents/seo-metadata-generator
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

### Generated SEO metadata

The generated title and description include character counts. A warning says title changes require an MDX rename, a URL change and redirect, and coordination with the documentation team.

[5:44](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=344s) [Generated meta title and description with character counts and a warning about changing the title.](https://assets.ai.engineer/sites/aiecode2025/sha256/b30870ac8d101efb4ae32487b24ebe01bc882dbc5ab5c9ec0f3f8d90e4bb23b5.webp)

Generated SEO Metadata
Total options: 1

Important: Title Change Implications
In the Docs repository, changing the meta title will:
• Replace the HTML title tag in the document
• Require renaming the MDX file to match the new title
• Change the page URL
• Require adding a redirect from the old URL to the new URL
Please coordinate with the documentation team before implementing title changes.

Meta Title
Navigate Twilio's Recording and Transcription Limits
Character count: 55 (Optimal length)

Meta Description
Discover key limitations of Twilio's transcription services, including compliance, search indexing, and media handling for better decision-making.
Character count: 157 (Optimal length)

Add context for new options
Generate New Options

Interface text:
docs-ai-buddy.vercel.app/agents/seo-metadata-generator
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

### Alt-text generator input

The alt-text generator accepts image uploads or a live Twilio docs URL. The presenter prepares to demonstrate generating alt text for multiple images.

[5:54](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=354s) [Alt-text generator with image upload and documentation URL options.](https://assets.ai.engineer/sites/aiecode2025/sha256/26e53646e628b12bcdb10b65103aa82c5b0163515e834efff3f88d6b35bcd2e1.webp)

Image Alt Text Generator

1. Provide the images that you want to generate alt text for. Click Choose files and select the images you want to use. Alternatively, paste a live Twilio docs URL.
2. Click Generate alt text.
3. Review the alt text in the Results section. If you like it, copy it into your .mdx file and make edits as needed. For details about syntax, see How to add images to docs pages.

Browse Files
Supported formats: image/*

OR

Paste Twilio Docs url here. For example, https://www.twilio.com/docs/messaging/tutorials/send-messages-with-messaging-services

Generate alt text

Interface text:
docs-ai-buddy.vercel.app/agents/alt-text-generator
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

### Images ready for alt text

The presenter has uploaded three images and is about to generate alt text for all of them. The form also accepts a live Twilio docs URL.

[6:05](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=365s) [Alt-text generator with three images uploaded and generation enabled.](https://assets.ai.engineer/sites/aiecode2025/sha256/fd4a1853bc3a9e2a9c075aa6239bc0cf8f33275ffd72b4b181d41870c4e71360.webp)

Image Alt Text Generator

1. Provide the images that you want to generate alt text for. Click Choose files and select the images you want to use. Alternatively, paste a live Twilio docs URL.
2. Click Generate alt text.
3. Review the alt text in the Results section. If you like it, copy it into your .mdx file and make edits as needed. For details about syntax, see How to add images to docs pages.

3 files uploaded:
generative-custom-operators.png
generative-operat-rs-training-examples-console.png
generative-operators-json-console.png

Browse Files
Supported formats: image/*

OR

Paste Twilio Docs url here. For example, https://www.twilio.com/docs/messaging/tutorials/send-messages-with-messaging-services

Generate alt text

Interface text:
docs-ai-buddy.vercel.app/agents/alt-text-generator
Docs AI Buddy
Overview and Release Notes
Automated Editor
Alt Text Generator
SEO Metadata Generator
Jargon Simplifier
Docs Outline Builder (coming soon)
Twilio
SS64 Command li...
Dashboard
Wagtail - Editing S...
Docs Dashboard v...
[DEVEDSB-171] 5-...
Mail Scanning, Virt...
About Us | First Sa...
Lifecycle of a Task...
All Bookmarks

### Guard-Rails & Iteration

Vale lint and CI tests help mitigate hallucinations; dataset tests and prompt audits address bias. PR reviews and Slack feedback help the team address misalignment and continually tune prompts.

[7:20](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=440s) [A diagram connects hallucination, bias, and stakeholder misalignment with AI and lists Vale lint, dataset tests, and human pull-request review.](https://assets.ai.engineer/sites/aiecode2025/sha256/0841f785eeb63d9b891e2c0bbb0b2af23e11eab9907b33f270f279459eb6ab76.webp)

Guard-Rails & Iteration

HALLUCINATION
BIAS
STAKEHOLDER MISALIGNMENT

AI

VALE LINT
DATASET TESTS
PR HUMAN REVIEW

2025 TWILIO INC. ALL RIGHTS RESERVED

### Your Roadmap

Identify a pain that hurts throughput, pick one repeatable, rule-based task, and loop with users at least weekly. Ship, measure, and refine.

[8:10](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=490s) [Three-step roadmap: identify pain, pick a single task, and loop with users.](https://assets.ai.engineer/sites/aiecode2025/sha256/f463530c92a3df884faef694272ee8ba24a15544859bab97ea8ad1c7725501ce.webp)

Your Roadmap

1
Identify Pain

2
Pick Single Task

3
Loop With Users

2025 TWILIO INC. ALL RIGHTS RESERVED

### Thank You

The speaker thanks the event team and attendees, and recognizes Twilio manager Carly Vanderwert and the leadership team.

[8:33](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=513s) [Thank you displayed inside a red outlined shape.](https://assets.ai.engineer/sites/aiecode2025/sha256/10351c392354c6ae87fe6cb3c086df456baceebf9ee3eebfb9c597652028a870.webp)

THANK YOU

2025 TWILIO INC. ALL RIGHTS RESERVED

## Full transcript

[0:00](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=0s) Hello, everyone. I'm Elmer Thomas, Principal Developer Educator at Twilio. Welcome to the AI Engineer World's Fair. Our talk today is called The Robots Are Coming For Your Job, And That's Okay.

[0:11](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=11s) Don't worry, the title is tongue-in-cheek. I'm here to show you why AI isn't something to fear, it's something we can embrace to work smarter. In the next few minutes, you'll see how our relatively tiny docs team uses AI agents to turbocharge workflows instead of trying to replace ourselves.

[0:27](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=27s) We'll look at the pain points, the agents we built, and the guardrails that keep us accurate. I've been a developer and builder for decades. Trends come and go. AI is the latest big one.

[0:37](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=37s) Like many of you, I wondered how it would change our day-to-day work. Today, you'll see what we've learned, warts and all. My colleague, Maria Bermudez, a lead developer of our AI docs buddy and resident backlog-busting coffee fanatic, will drive a live agent walkthrough, so you can judge the results yourself.

[0:55](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=55s) Here's a game plan. We'll cover the pain, our agent architecture, Maria's demo, and a roadmap that you can steal for your own team. Let's dive in. Picture this, a tiny docs team, a flood of Jira tickets.

[1:11](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=71s) Three headaches jump out. Number one, error-prone first drafts from hundred-plus product teams. Two, time sink grooming, like style checks, alt text, SEO. Three, hallucination risk if we let GenAI run wild.

[1:25](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=85s) We needed leverage, not burnout. Instead of one megabot, we built six single-purpose agents behind a simple Next.js front end. The automated edum- editor fixes grammar, formatting, and accuracy.

[1:42](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=102s) Image alt text gen- text generator gives you instant accessibility wins. The jargon simplifier turns dev speak into plain English. SEO metadata provides title and description and character, keeping it character count safe.

[1:58](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=118s) The docs outline builder recommends nav and structure. That's coming soon. Super excited about that. Our Slack back bot helps triage help channel requests. Each agent tackles a repetitive, well-scoped job, so humans can focus on judgment and clarity.

[2:16](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=136s) Rule of thumb, pick tasks that are repeatable, high volume, and low creativity. That's the sweet spot for an AI helper. Here's the flow behind every request. We have a Next.js UI that feeds into a custom GPT-4-01 agent.

[2:39](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=159s) We use the appropriate model for the particular job. The custom GPT is, it has our style guide, and it has our rubric that is baked in. It's actually retrieved from an Airtable so that folks can easily collaborate on it.

[2:56](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=176s) We have a validation layer that includes Vale linting and CI/CD tests. The GitHub PR adds code owner review and makes it much easier for us to scrutinize the changes that the agents are suggesting.

[3:12](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=192s) A human hits the merge only when it's right, and usually, we have built into their product and engineering reviews as well. So we have several human eyes before we actually consider hitting that merge button.

[3:29](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=209s) This layered approach, it crushes hallucinations without slowing us down. Maybe not crushes them, but it reduces them significantly. All right. It's time to prove this isn't vaporware. I'm handing the screen to Maria Bermudez, the brains behind much of the AI docs buddy code.

[3:49](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=229s) Maria, take it away.

[3:51](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=231s) Thank you, Elmer. So real quick, I wanna show you how we use Copilot to auto-generate our overview and release notes page. And from here, you can get into any of the agents that you want to work with.

[4:02](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=242s) So I'm gonna start with the automated editor. One cool thing about this is you not only can load an MDX file, uh, a Markdown file, but you can also plug in a live URL, and I'm gonna do that option right now.

[4:16](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=256s) This does take a little while, so we added this graphic here because we're using the o1 model. We've done a lot of experimentation and figured out this is the best model to consistently apply our style guide and rubric.

[4:31](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=271s) And here you go. And then on the...

[4:35](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=275s) It shows you the diff of what it changed, and, uh, in the changes made tab, you actually have a list of, um, explanation of the original text, the revised text, and what it changed and what style guidance and rubric item the changes pertain to.

[4:53](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=293s) Um, it's not perfect, as, as you can see. Like, um, sometimes it catches that there's a missing SEO description, but other times it doesn't. So luckily, we have an SEO metadata generator, and I'm going to plug in this page into that one.

[5:12](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=312s) So what I'm actually gonna do is I'm going to upload the file

[5:18](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=318s) and then go back to Conversational Intelligence. So now I'm going to generate SEO metadata for that same page that we just edited.

[5:41](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=341s) And there you go. So it generates the meta titles. You have to specify if you don't want that, and then it gives you the meta description, um, accounting for character limitations.

[5:52](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=352s) And now I'm gonna show you the alt text generator.

[5:56](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=356s) It has the same option that we have, where you can plug in a live URL, but I'm gonna show you, um, picking multiple pages and

[6:06](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=366s) Having our agent generate alt text for all of those images. So I'm gonna click on Generate alt text now.

[6:16](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=376s) And you see how quickly that got done, and it also conforms to the format that we require from our docs platform.

[6:25](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=385s) And lastly, I'm gonna show you the jargon simplifier. So you can take whatever, like, pages that you're reviewing. This is really helpful, not only for writing, but when we're re-reviewing a pull request, and it gives us, like, a good way to provide quick edits to those pages.

[6:43](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=403s) So I'm gonna plug in a text that I've prepared, and I'm going to click on Simplify.

[6:56](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=416s) And there you go. It also gives you a diff like our n- our editor. And on the next tab, you have the revised text that you can quickly copy and apply as a pull request, uh, comment or just edit the file directly.

[7:12](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=432s) And that's it. We're working on making our agents talk to each other eventually, but we, we have a lot of work to do on that front.

[7:20](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=440s) Thank you, Maria. That was wonderful. Now, tools alone don't guarantee quality. Our guardrails tackle three big risks. Hallucinations, we use tools like Vale Lint and CI Test to help mitigate that in combination with the various eyes of, of human stakeholders.

[7:41](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=461s) Bias, dataset tests and prompt audits are the way to tackle those issues. Stakeholder misalignment, so weekly PR reviews. Sometimes these are compressed into days and sometimes hours.

[7:57](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=477s) And we also have Slack feedback loops, generally with the product managers and engineering team. These feedback cycles, they let us tune the prompts continuously instead of hoping the model magically stays perfect.

[8:12](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=492s) S- Now, please steal this three-step playbook. Identify one pain that's killing throughput. Two, pick a single task that's repeatable and rule-based. And then three, loop with your users weekly at least.

[8:27](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=507s) Ship, measure, and refine. Stack a few of these wins and watch your team's velocity jump. Thanks again to the AI Engineer World's Fair team for hosting us, and thanks to you for spending time with us today.

[8:42](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=522s) Big shout out to Carly Vanderwert, our manager at Twilio, who keeps encouraging us to push the boundaries with AI, along with the rest of our leadership. We appreciate your attention.

[8:53](https://www.youtube.com/watch?v=pSqpC7fFLZA&t=533s) Now go build something amazing with AI.
