Request // Response

Request // Response Trailer Bonus Episode 2 Season 1

API design ergonomics, coding on airplanes, and the true signal of good DevEx | Robert Ross (CEO FireHydrant)

API design ergonomics, coding on airplanes, and the true signal of good DevEx | Robert Ross (CEO FireHydrant)API design ergonomics, coding on airplanes, and the true signal of good DevEx | Robert Ross (CEO FireHydrant)

00:00
Robert Ross (@BobbyTables) is the CEO of FireHydrant.

We discuss the journey of building FireHydrant, the evolution of API design, and the impact of gRPC and REST on developer experience.

We also talked about the role of LLMs in API design, the shift towards data consumption trends in enterprises, and how great developer experience is measured by a simple litmus test.


Listen On
Apple Podcasts | Spotify


Subscribe to Request // Response
If you enjoyed this podcast, you can be the first to hear about new episodes by signing up at https://speakeasy.com/post/request-response-robert-ross


Show Notes
[00:00:00] Introduction
- Overview of discussion topics: building FireHydrant, gRPC, API design trends, and LLMs in APIs.

[00:00:42] The Story Behind FireHydrant
- Robert’s background in on-call engineering and why he built FireHydrant.
- The problem of incidents and automation gaps in on-call engineering.

[00:02:16] APIs at FireHydrant
- FireHydrant’s API-first architecture from day one.
- Moving away from Rails controllers to a JavaScript frontend with API calls.
- Today, over 350 public API endpoints power FireHydrant’s frontend and customer integrations.

[00:03:50] Why gRPC?
- Initial adoption of gRPC for contract-based APIs.
- Evolution to a REST-based JSON API but with lessons from protocol buffers.
- Would Robert choose gRPC today? His thoughts on API design best practices.

[00:06:40] Design-First API Development
- The advantages of Protocol Buffers over OpenAPI for API design.
- How API-first development improves collaboration and review.
- Challenges with OpenAPI’s verbosity vs. Protobuf’s simplicity.

[00:08:23] Enterprise API Consumption Trends
- Shift from data-push models to API-first data pulls.
- Companies scraping API every 5 minutes vs. traditional data lake ingestion.
- FireHydrant’s most popular API endpoint: Get Incidents API.

[00:10:11] Evolving Data Exposure in APIs
- Considering webhooks and real-time data streams for API consumers.
- Internal FireHydrant Pub/Sub architecture.
- Future vision: "Firehose for FireHydrant" (real-time streaming API).

[00:12:02] Measuring API Success (KPIs & Metrics)
- Time to first byte (TTFB) as a key metric.
- API reliability: retry rates, latency tracking, and avoiding thundering herd issues.
- The challenge of maintaining six years of stable API structure.

[00:17:12] API Ergonomics & Developer Experience
- Why Jira’s API is one of the worst to integrate with.
- The importance of API usability for long-term adoption.
- Companies with great API design (e.g., Linear).

[00:18:14] LLMs and API Design
- How LLMs help in API design validation.
- Are LLMs changing API consumption patterns?
- Rethinking API naming conventions for AI agents.

[00:22:02] Future of API Ergonomics for AI & Agents
- Will there be separate APIs for humans vs. AI agents?
- How agentic systems influence API structures.
- The need for context-rich naming conventions in APIs.

[00:24:02] What Defines Great Developer Experience?
- The true test of developer experience: Can you be productive on an airplane?
- Importance of great error messages and intuitive API design.
- The shift towards zero-docs, self-explanatory APIs.


More Quotes From The Discussion

Enterprise Data Teams are Now More Receptive to APIs

"I think more and more people are willing to pull data from your API. There used to be kind of a requirement many years ago of like, you need to push data to me, into my data lake. And the problem with that is, sure, we can push data all day long, but it may not be in the format you want.

It may not be frequent enough. Like, you might have stale data if we're doing, a nightly push to our customers' data lakes.

So, a lot of our customers are scraping our API every five minutes in some cases to get their incident data. And what our customers tell us is that they have data analyst teams. They have data analytics engineers that their job is to write scripts that pull the data into their internal data lake from all of their vendors. And I think that's, I don't know if it's a new trend or the new normal, I'm not quite sure, but it's definitely something I've noticed is that enterprises are spending more time building the teams to get the data in the exact format, the exact time that they want it, as opposed to having the requirement of push the data to me."

On LLMs and API Design

“I don't think that LLMs are going to fundamentally change how we interact with APIs. At least not in the short term.

I think that they'll help us build better clients and I think they'll help us build better APIs, but the moment we start using them, I mean, that's just going to be code calling code. I just don't think that LLMs are going to get too integrated at that point.”


Referenced
- FireHydrant the end-to-end incident management platform (https://firehydrant.com)
- Linear API docs (https://linear.app/docs/api)
- Offset's Materialize RBAC system (https://materialize.com/docs/manage/access-control/rbac)
- Speakeasy's Terraform generator (https://docs.speakeasyapi.dev/docs/using-speakeasy/client-sdks-generation/terraform)


Production by Shapeshift | https://shapeshift.so

For inquiries about guesting on Request // Response, email samantha.wen@speakeasy.com.

What is Request // Response?

Interviews with developers and API technology leaders.
Hosted by Sagar Batchu, CEO of Speakeasy.

speakeasy.com