On The Graph with Apollo

On The Graph with Apollo

By Dan Boerner & Apollo GraphQLBusinessTechnology
Download on the App Store

All On The Graph with Apollo episodes

  • Ep. 5 - Stop Chasing the Perfect Schema | Arne Wieding, Delivery Hero

    Arne Wieding wanted two perfect things at Delivery Hero: the perfect schema, and one graph for the whole company. He let go of both, and in this episode he explains why.

    Delivery Hero is one of the biggest names in food delivery outside the US, with about a billion orders a quarter in around 65 countries. During the COVID boom, its teams, services and APIs multiplied until building one feature meant aligning across too many teams. Arne, a Principal Software Engineer there, helped answer that with a GraphQL schema that became a shared vocabulary for engineers from Berlin to Singapore.

    We get into why one graph for everything stopped being the goal, bringing REST APIs in without new code, server-driven tiles that move a discount badge in minutes instead of a mobile release, a bot that reviews schema changes, and why GraphQL and MCP fit so well for LLMs.

    Stay to the end for his advice: stop chasing the perfect schema, and skip the big bang migration.

    Arne Wieding, Principal Software Engineer, Delivery Hero: https://www.linkedin.com/in/arnewieding/
    Dan Boerner, host: https://www.linkedin.com/in/dboerner/

    Chapters
    0:00 A perfect fit for LLMs
    0:21 Introduction
    2:14 When growth slowed every team down
    3:32 The schema as a shared vocabulary
    4:27 Designing the API before the code
    6:48 One graph for all of Delivery Hero?
    8:52 Bringing in REST APIs without new code
    11:28 Should the API decide what the app shows?
    13:44 From a mobile release to minutes
    14:51 Keeping every team's API consistent
    16:19 A bot that reviews schema changes
    18:55 Ask for the query, skip the docs
    20:44 Not the only evangelist anymore
    21:54 His advice: no perfect schema, no big bang

    If you're working to power your AI efforts and APIs are in the way, email me and I'll connect you with other API champions like Arne: [email protected]

    Watch the full episode on YouTube: https://youtu.be/SLgmMBhCev4

    On The Graph is a show from Apollo GraphQL, hosted by Dan Boerner, about the people who build the layer everything else runs on.

    26 min
  • Ep. 4 - No one cares what Booking.com's API runs on. It's GraphQL. | Chris Ernst

    Each of the billion room nights booked on Booking.com every year runs through a GraphQL API layer that handles over 200,000 requests per second. Yet their apps, underlying platform services and now new agentic experiences are driving constant change and evolution. How do they keep up with all that change and scale?

    In this episode I talk with Chris Ernst, Senior Software Engineer on Booking.com's API platform team, about exactly that. When he joined, three people approved every schema change by hand. Then the API grew 25 times over, and for a stretch nobody was really checking. Now LLMs review schema changes as teams build, and make up realistic test data so front-end teams don't have to wait for the back end.

    Chris shares how a stable GraphQL contract let them safely modernize their original monolith into a federated graph without ruining anyone's vacation, why mock data is where LLM hallucination becomes a feature, what AI agents get from an API built for people, and why he'd still build a graph today.

    Stay to the end for his advice to teams starting out: just go for it, and invest heavily in developer experience.

    Chris Ernst, Senior Software Engineer, Booking.com: https://www.linkedin.com/in/christian-ernst11/
    Dan Boerner, host: https://www.linkedin.com/in/dboerner/

    Chapters
    0:00 No one really cares
    0:10 Introduction
    1:11 A billion room nights, 250,000 requests a second
    2:38 Mapping the legacy rules no one person knows
    5:33 A steady contract, and a platform that never stops changing
    8:10 Three people approving every schema change
    9:33 The YOLO period, and the reviews that replaced it
    11:57 Mock data, where hallucination becomes a feature
    14:27 What agents get from a typed, connected graph
    16:27 Where MCP fits today: a first stepping stone
    18:39 Making the case, from developer to executive director
    21:16 The job's not finished yet
    22:13 Starting over today: would he still build a graph?
    23:29 Advice for teams starting out

    If you're working to power your AI efforts and APIs are in the way, email me and I'll connect you with other API champions like Chris: [email protected]

    Watch the full episode on YouTube: https://youtu.be/T8WnOgu7JgU

    On The Graph is a show from Apollo GraphQL, hosted by Dan Boerner, about the people who build the layer everything else runs on.

    25 min
  • Ep. 3 - Starbucks runs on a graph. Now agents want in. | Sharon Gorla

    Since 2016, a lot of what happens between tapping the Starbucks app and grabbing your drink has run through a graph API. Then leadership changed, priorities changed, and parts of the platform went back to a monolith. What survived wasn't the implementation. It was the contract.

    In this episode I talk with Sharon Gorla, Lead Engineer and Architect at Starbucks. We cover what happened when the org and priorities changed, why she jokes that GraphQL was accidentally built for AI, and what a decade of platform work taught her about people.

    We get into why AI agents are a new kind of consumer, but you don't need a new platform for them, and why every technical champion needs an organizational one.

    Sharon Gorla, Lead Engineer and Architect, Starbucks: https://www.linkedin.com/in/sgorla/
    Dan Boerner, host: https://www.linkedin.com/in/dboerner/

    Chapters
    0:00 The contract survived
    0:26 Introduction
    2:10 Agents are a new kind of consumer
    5:00 Did GraphQL already have its moment?
    6:14 Accidentally built for AI?
    9:34 Clients come and go
    10:36 Making the case inside Starbucks
    13:44 A kiosk team that needed no meetings
    14:36 Leadership changed. The contract didn't.
    18:20 Do agents need a new platform? Where MCP fits
    19:51 Tokens, and the bill coming due
    21:59 Champion people, not just technology

    If you're working to power your AI efforts and APIs are in the way, email me and I'll connect you with other API champions like Sharon: [email protected]

    Watch the full episode on YouTube: https://youtu.be/dwHTaeVQScc

    On The Graph is a show from Apollo GraphQL, hosted by Dan Boerner, about the people who build the layer everything else runs on.

    28 min
  • Ep. 2 - From green screen to AI agents | Praga Prakasam

    American Airlines turns 100 this year and still runs on Sabre, a reservation system from the 1960s. Now leadership wants new AI agents shipped every 90 days.

    Praga Prakasam was ready. Years before anyone was saying “agents,” he'd put a graph API in front of all of it. He's a Senior Principal Architect and 13-year veteran at American, where he owns the API platform and was recently tapped to speed up its customer-facing AI agents.

    We get into why probabilistic agents need deterministic answers underneath, why an agent that works on your laptop isn't ready for production, how he stops an agent from canceling your reservation, and why putting MCP straight on top of your existing APIs repeats a mistake teams already made once.

    And listen for what he says when I ask how it feels to see a bet made years early finally pay off.

    Praga Prakasam, Senior Principal Architect, American Airlines: https://www.linkedin.com/in/pragaprakasam/
    Dan Boerner, host: https://www.linkedin.com/in/dboerner/

    Chapters
    0:00 Cold open
    0:29 Introduction
    2:28 Praga, Sabre, and 100 years of American Airlines
    3:41 Modernizing around a system you don't control
    6:14 What changed for the traveler experience
    8:51 Then the AI moment arrived
    10:51 Why they tapped an API architect to lead AI efforts
    13:56 Agents need deterministic answers
    15:40 Shipping agent tools without writing new code
    16:17 The MCP mistake almost everyone makes
    21:05 What a long bet feels like when it pays off
    24:25 Why security is the part that ships it

    If you're working to power your AI efforts and APIs are in the way, email me and I'll connect you with other API champions like Praga: [email protected]

    Watch the full episode on YouTube: https://youtu.be/_dLUOoCnqsw

    On The Graph is a show from Apollo GraphQL, hosted by Dan Boerner, about the people who build the layer everything else runs on.

    29 min
  • Ep. 1 - What if everything goes right? | Dhanik Alkegama

    N-able's customers kept asking for something they couldn't name: a way through the API sprawl left by years of acquisitions. Dhanik Alkegama's team built them a graph, and it caught on faster than anyone planned. Then Apollo MCP Server arrived, and the thing customers could never describe became something they could simply ask their agents for.

    In this episode I talk with Dhanik Alkegama, Engineering Manager at N-able, where the graph sits in front of products protecting over 9 million endpoints. We cover his push to make education a requirement rather than a suggestion, why he thinks governance can speed teams up instead of slowing them down, and why agents arrived to find the graph already waiting.

    Since we recorded, N-able's MCP server has gone public, and their partner Halo ITSM built their own MCP against it, unprompted.

    Dhanik Alkegama, Engineering Manager, N-able: https://www.linkedin.com/in/dhanik-alkegama/
    Dan Boerner, host: https://www.linkedin.com/in/dboerner/

    Chapters
    0:00 “They were praying for a graph”
    0:25 When tech adoption outruns the plan
    2:21 N-able API mess
    7:40 Making the case to leadership
    9:18 Planning for one team. A dozen showed up
    12:26 Success becomes the problem
    13:43 Education as Infra
    16:03 Four people governing a dozen teams
    20:33 Apollo MCP Server arrives just in time
    26:20 Contracts: one graph serves two audiences
    27:42 Three lessons Dhanik would pass on

    If you're working to power your AI efforts and APIs are in the way, email me and I'll connect you with other API champions like Dhanik: [email protected]

    Watch the full episode on YouTube: https://youtu.be/OJEX7ANnxRo

    On The Graph is a show from Apollo GraphQL, hosted by Dan Boerner, about the people who build the layer everything else runs on.

    30 min