Latest / The Augur Dispatch / Anthropic's engineer academy and the hidden cost of switching AI vendors

Transcript

Anthropic's engineer academy and the hidden cost of switching AI vendors · The Augur Dispatch

• 0:06
  SPEAKER_00: An engineer comes back from vendor training able to make your AI system work.
• 0:11
  SPEAKER_00: That's useful.
• 0:12
  SPEAKER_00: The question to ask before the next contract is whether somebody else could understand what they built and move it to a different supplier.
• 0:19
  SPEAKER_00: Take the training if it helps.
• 0:21
  SPEAKER_00: Keep the deployment knowledge inside your company and test whether the handover makes sense to someone who wasn't in the room.
• 0:27
  SPEAKER_00: This is the Augur Dispatch for Saturday, October 3rd, 2026.
• 0:32
  SPEAKER_00: Anthropic has committed$100 million to Claude Frontier Academy, aiming to train 10,000 deployment engineers by the end of 2027.
• 0:42
  SPEAKER_00: These engineers connect models to the systems and working routines that companies already have.
• 0:47
  SPEAKER_00: The first participants include people from consulting firms such as Accenture and Deloitte, alongside customer organizations, including Morgan Stanley and Commonwealth Bank of Australia.
• 0:58
  SPEAKER_00: Their employers nominate them.
• 1:00
  SPEAKER_00: They return to a Claude project at their own organization.
• 1:03
  SPEAKER_00: That matters because this isn't simply anthropic sending its own employees into somebody else's office.
• 1:09
  SPEAKER_00: It's also teaching the customers people and the consultants that customers hire how to build around Claude.
• 1:16
  SPEAKER_00: Picture an operations manager whose new AI workflow is finally running.
• 1:21
  SPEAKER_00: The consultant knows why one data feed needs special treatment, which exceptions need a person, and what breaks at the end of the month.
• 1:28
  SPEAKER_00: The manager has a working service.
• 1:30
  SPEAKER_00: Whether the company owns enough knowledge to change it is a separate question, and a successful launch doesn't automatically supply the answer.
• 1:38
  SPEAKER_00: Anthropic's program makes that knowledge concrete.
• 1:41
  SPEAKER_00: It starts with an intensive, including a graded practical exercise, then a 12-week residency leading a real claw deployment with support from Anthropic engineers.
• 1:51
  SPEAKER_00: There's another assessment at the end.
• 1:54
  SPEAKER_00: Anthropic expects the first final credentials in early 2027.
• 1:58
  SPEAKER_00: It calls this the Academy's first program, with additional programs planned for next year.
• 2:03
  SPEAKER_00: That is useful training.
• 2:05
  SPEAKER_00: It is also repeated practice in solving business problems with one vendor's technology.
• 2:10
  SPEAKER_00: My expectation is that engineers trained this way will be more likely to reach for Claude on their next project.
• 2:16
  SPEAKER_00: That's an inference about habits and switching costs, not a measured finding about the academy's graduates.
• 2:22
  SPEAKER_00: It doesn't establish bad intent or mean that learning one platform prevents an engineer from learning another.
• 2:29
  SPEAKER_00: If the trainee is your employee, the knowledge can stay in your business.
• 2:33
  SPEAKER_00: If the trainee works for a consultancy, you need a handover that transfers more than the finished application.
• 2:39
  SPEAKER_00: A folder named Final Documentation is a promising start.
• 2:43
  SPEAKER_00: It isn't proof that anyone can use it.
• 2:46
  SPEAKER_00: The financial context explains why I think this training deserves attention beyond the education budget.
• 2:52
  SPEAKER_00: Big Technologies analysis and TechCrunch's reporting put planned infrastructure spending over the next decade at$518 billion, with roughly four-fifths non-cancelable, according to Big Technologies account.
• 3:05
  SPEAKER_00: Those figures come through reporting on the prospectus, the document prepared before a public share sale.
• 3:11
  SPEAKER_00: I haven't independently audited those commitments, and they aren't all payable tomorrow.
• 3:17
  SPEAKER_00: Still, a bill that cannot be canceled creates an incentive to keep customers for longer, even when the company is already growing quickly and bringing in much more revenue.
• 3:26
  SPEAKER_00: Against that scale, a$100 million training commitment is small.
• 3:31
  SPEAKER_00: It could help Anthropic build a much larger pool of people who know how to make cloud valuable inside a business.
• 3:37
  SPEAKER_00: That doesn't prove the Academy will reduce customer departures.
• 3:41
  SPEAKER_00: It explains why I read the program as an investment in keeping business as well as in developing talent.
• 3:47
  SPEAKER_00: OpenAI is competing for the same work.
• 3:50
  SPEAKER_00: IBM's partnership includes OpenAI technology in its consulting platform, a dedicated practice, and specialized engineers and consultants trained through OpenAI's partner network.
• 4:02
  SPEAKER_00: The customer benefit can be substantial.
• 4:04
  SPEAKER_00: In an OpenAI case study, Chatham Financial says an application built with Codex cut trade review from about 30 minutes to under 4.
• 4:13
  SPEAKER_00: The application gathers transaction evidence, compares terms, and flags discrepancies.
• 4:18
  SPEAKER_00: Chatham describes the timing as an early measurement and says it is comparing the application's results with experienced reviewers before expanding automation.
• 4:28
  SPEAKER_00: That qualification belongs beside the speed figure.
• 4:31
  SPEAKER_00: A faster review becomes valuable when the person responsible for a transaction can still trust the result and trace the evidence behind it.
• 4:39
  SPEAKER_00: For your next rollout, start with one workflow you already understand.
• 4:43
  SPEAKER_00: Ask the technical lead to separate the integration work done by employees from work done by vendor engineers and vendor-trained contractors.
• 4:51
  SPEAKER_00: Then ask what would need rebuilding if the model changed.
• 4:55
  SPEAKER_00: Included the tests and operating knowledge, not just the connection to the model.
• 5:00
  SPEAKER_00: Treat that as an estimate of part of the switching cost, rather than a complete exit bill.
• 5:05
  SPEAKER_00: Here's a practical handover test.
• 5:07
  SPEAKER_00: Give the documentation to an internal engineer who didn't build the system and ask them to explain its failure handling and reproduce a small, safe part of the workflow.
• 5:17
  SPEAKER_00: Make the technical lead responsible for reviewing the gaps before the next expansion decision.
• 5:22
  SPEAKER_00: If the handover fails, pause expansion until somebody fixes it.
• 5:26
  SPEAKER_00: Buying a second model won't fill in a missing explanation.
• 5:30
  SPEAKER_00: There is more room to choose suppliers in another corner of AI, models that return a decision rather than a paragraph.
• 5:38
  SPEAKER_00: Imagine a support lead trying to route incoming tickets.
• 5:41
  SPEAKER_00: The useful output may simply be billing, technical support, or account access.
• 5:46
  SPEAKER_00: A beautifully written essay would just give the queue something else to read.
• 5:50
  SPEAKER_00: Olama, software for running models locally, already supported decision models from Bespoke Labs and Together AI in its initial release of this feature.
• 5:59
  SPEAKER_00: Cloudflare's Clef and Clef Flash now add another supplier through the same interface.
• 6:05
  SPEAKER_00: An interface is the agreed way one piece of software asks another for something.
• 6:10
  SPEAKER_00: Sharing it can reduce some of the work of trying a different model, although it doesn't make their judgments equally reliable.
• 6:16
  SPEAKER_00: Cloudflare's models also accept images in OLAMA.
• 6:20
  SPEAKER_00: The cited release of Llama.cpp, another local runtime, supports Clef only for text so far.
• 6:26
  SPEAKER_00: I'd pilot this for internal ticket routing, with the support lead owning the result.
• 6:31
  SPEAKER_00: Compare it with the current routing process on previously reviewed tickets and keep people in control of anything consequential.
• 6:38
  SPEAKER_00: At the end of two weeks, review routing errors and handling time.
• 6:42
  SPEAKER_00: Stop if serious misroutes increase.
• 6:45
  SPEAKER_00: Expand only if quality holds and the operational saving is worth maintaining the new system.
• 6:51
  SPEAKER_00: Don't standardize the whole company on the format yet.
• 6:54
  SPEAKER_00: Competing interfaces, including OpenAI's Decisions API, leave the longer-term choice unsettled.
• 7:01
  SPEAKER_00: A smaller model can also change the economics of a repeated task.
• 7:05
  SPEAKER_00: AI2's Astabrief writes cited scientific reports from a research question and literature excerpts retrieved beforehand.
• 7:13
  SPEAKER_00: A researcher checking a narrow question may value a useful preliminary report sooner, provided the citations really support the claims.
• 7:21
  SPEAKER_00: A shorter wait is less helpful if it just moves the work into correcting the answer.
• 7:26
  SPEAKER_00: AI2 reports about 51 seconds for fast mode across its full pipeline, compared with about three minutes for claw-powered thinking mode.
• 7:34
  SPEAKER_00: That's roughly three and a half times faster.
• 7:37
  SPEAKER_00: The separate claim of nearly 10 times faster concerns report generation against the proprietary models it tracked.
• 7:44
  SPEAKER_00: It shouldn't be substituted for the full pipeline comparison.
• 7:47
  SPEAKER_00: The lab also published quality evaluations.
• 7:50
  SPEAKER_00: Much of that work was done in 2025, and it says it hasn't rerun the full evaluation against today's frontier models.
• 7:58
  SPEAKER_00: For a research team, I'd ask the research lead to compare both approaches on its own recurring questions for two weeks, measuring citation accuracy and reviewer time as well as speed.
• 8:09
  SPEAKER_00: Stop if important claims lose their support.
• 8:12
  SPEAKER_00: These are published results from the lab, not proof that a small model will match your current system.
• 8:18
  SPEAKER_00: They are a reason to test a particular job, not replace everything.
• 8:22
  SPEAKER_00: One dated call today.
• 8:24
  SPEAKER_00: I give a 40% chance that, by December 31st, 2027, Anthropic publicly states that at least 10,000 engineers have completed Claude Frontier Academy training.
• 8:35
  SPEAKER_00: A hit requires an Anthropic blog post, press release, regulatory filing, or academy page published by that date reporting at least 10,000 completed trainees.
• 8:45
  SPEAKER_00: Without that statement by the deadline, it's a miss.
• 8:48
  SPEAKER_00: Enrollment and the early badge don't establish completion of the full residency.
• 8:53
  SPEAKER_00: The first published completion counts are the evidence to watch.
• 8:57
  SPEAKER_00: Augur's calls are dated predictions with explicit tests tracked on the website so the outcome can be checked later.
• 9:04
  SPEAKER_00: For now, adopt the knowledge handover discipline.
• 9:07
  SPEAKER_00: Pilot the narrow alternatives where you can check quality and watch whether the academy delivers on its goal.
• 9:13
  SPEAKER_00: The written brief with every source is at AugurDispatch.ai.
• 9:17
  SPEAKER_00: Portions of the Augur Dispatch are AI generated.
• 9:20
  SPEAKER_00: That is the Augur Dispatch for Saturday, October third, twenty twenty six.
• 9:25
  SPEAKER_00: Thank you for listening.