# Augusto Marietti

Augusto Marietti shares Kong’s work on API infrastructure and AI connectivity, including AI gateways, token costs, and semantic routing.

- Profile: https://b2binfluencer.marketing/creators/augusto-marietti/
- LinkedIn: https://www.linkedin.com/in/sonicaghi
- Topics: artificial-intelligence, ai-governance
- Based on 14 dated posts, latest 2026-10-05

## Summary

Augusto Marietti explains API infrastructure, AI gateways, semantic routing, and token costs for technical and finance audiences.

- Best for: A B2B infrastructure or AI platform campaign reaching enterprise CFOs and technical buyers.
- Not for: Consumer-facing lifestyle, retail, or app launches needing on-camera video and measured reach.
- Check first: Confirm current sponsor policies, partnership specs, disclosed examples, and inquiry routes.

### Takeaways

- Software is shifting to APIs, MCP tools, and CLI for AI agents.
- Semantic routing sends simple high-volume requests to cheaper models, complex reasoning to frontier models.
- Gateway metrics can map to P&L equivalents and show costs, margins, and spend drivers.

## Main ideas

- **As software shifts toward AI agents, APIs become the way in**: Marietti argues that software is being rebuilt for machines rather than people. Agents need programmatic interfaces—such as APIs, MCP tools, command-line interfaces and protocols—instead of screens designed for human users. They present API connectivity as a basic requirement for the next phase of software.
- **More AI use does not have to mean costs rising at the same rate**: Marietti challenges the assumption that token use and spending move in lockstep. They argue that an AI gateway can route requests by intent, matching routine tasks with cheaper models and reserving frontier models for complex reasoning. Routing choices can change the cost curve even as usage grows.
- **Make AI infrastructure costs legible to finance teams**: Marietti says technical measures become more useful in executive conversations when translated into financial consequences. They describe this as a way for CFOs and technical teams to discuss AI spending without expecting finance leaders to interpret engineering metrics on their own.

## Work to read

- [“Software was going headless” — the argument behind the API-first pitch](https://www.linkedin.com/posts/sonicaghi_noaiwithoutapis-activity-7450940202979807232-beKH)
- [The token-cost argument: using more tokens while spending less](https://www.linkedin.com/posts/sonicaghi_scary-thought-you-might-be-having-right-activity-7483554605994295296-_8EJ)
- [Translating engineering metrics into finance language for CFOs](https://www.linkedin.com/posts/sonicaghi_i-meet-with-more-cfos-than-i-used-to-we-activity-7498754986835714048--M92)
- [Returning to Milan: origin story, Kong Day and European R&D](https://www.linkedin.com/posts/sonicaghi_last-week-we-opened-our-kong-milan-office-activity-7488941810841559042-kxcT)
- [Hiring a CFO with public-market and $1B-scale experience](https://www.linkedin.com/posts/sonicaghi_im-excited-to-welcome-bruce-felt-to-the-activity-7430293650678452225-3sCO)
- [Paris office and European expansion](https://www.linkedin.com/posts/sonicaghi_the-incredible-kong-team-just-opened-our-activity-7481351490574401536-t_w-)
- [Inside the Summit agenda: token cost, governance and APIs for AI](https://www.linkedin.com/posts/sonicaghi_we-dont-always-have-a-ball-pit-at-summit-activity-7490758550458232832-xUeQ)

## Brand relationships

- **Kong** (owned_business)

## Partnerships

Across seven LinkedIn posts dated 19 February to 27 August 2026, Augusto Marietti discusses Kong’s API infrastructure, AI gateway, token cost management, semantic routing, office expansion, and Kong Summit. Kong Summit is described as an event with an executive forum, content tracks, and attendees, not as a creator sponsorship offering. Before planning a collaboration, ask for current sponsor policies, partnership specifications, disclosed examples, and inquiry routes.

## Alternatives

- Robert Ford: Choose Ford when the buyer's decision turns on the finance side of AI spend — when CFOs and controllers need to agree on what an AI programme should optimise, not how the plumbing routes requests. Their two posts work through a four-number framework that is unusually usable in a finance meeting: business value as the ceiling the business sets before anyone spends, marginal value as the method for deciding when to stop, ROI as a report that looks best when you spend least, and net value as the result worth steering by. They make the point concrete with a task example — spend about $15 on a task and keep roughly $38, spend about $32 and keep about $48 — and are explicit that the answer sits task by task, not only on the company bill. They also widen the cost definition beyond tokens to vector databases, caching layers and orchestration, which most AI cost conversations skip. When the need is AI cost governance as a discipline, or a speaker who can hold a CFO's attention for a full session, Ford is the closer match on the argument itself.
- Josh Cavalier: Choose Cavalier when the deliverable is a demonstration rather than an argument, or when the audience includes learning, talent and workforce leaders alongside technical buyers. Their AT&T write-up comes closest to the subject of AI gateway routing and cost trade-offs: they report the company processing about 45 billion tokens a day through an AI gateway that weighs accuracy, cost and latency and routes accordingly; open models saving 80–90% in some applications; managers holding token budgets as capital allocators; and the trade-off question of a 56% cost cut for a 2% quality drop. They then extend the discussion into what most infrastructure pitches omit — proprietary data, documented processes, and whether companies reinvest productivity gains in people. Their second post documents building a 48-second promo with an AI assistant in under an hour, naming exactly where the process broke down. It suits sponsors seeking someone who can show AI adoption in processes on camera or in a live build rather than just describe it.
- Sushil Barwal: Choose Barwal when the campaign targets mid-market operations or eCommerce rather than enterprise infrastructure buyers. Their posts are decision shortlists a team can act on the same week: five internal processes worth automating first — lead qualification and routing, tier-1 support and order inquiries, cross-system data synchronisation, document verification and client onboarding, and real-time operational monitoring — with the manual and automated versions of each stated plainly. The eCommerce post takes the same approach to site performance, naming the three architecture issues they check first and the fixes for each, then giving a full-range example of moving from 3.8 seconds to under 1.2 seconds. Use Barwal for a practical automation or performance agenda aimed at a mid-market audience, or when the sponsor wants content that reads as implementable rather than strategic.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/