# Amdahl: search and evals for GTM agents

Canonical: https://mudpie.ai/companies/amdahl/
Breadcrumb: [Home](https://mudpie.ai/) / [Companies](https://mudpie.ai/companies/) / [Amdahl: search and evals for GTM agents](https://mudpie.ai/companies/amdahl/)
Author: Ali Abouelatta (https://mudpie.ai/authors/ali-abouelatta/)
Published: 2026-09-19
Updated: 2026-09-19
Research type: Company profile
Method: Company and accelerator sources checked 2026-09-19. Product claims are attributed to their sources; this is research, not a hands-on product trial.

Amdahl is search and evaluation infrastructure for GTM agents. It fits revenue teams that want AI to use their own deal, call and CRM context—and want the output to come back with citations and a way to grade whether it was useful.

## What it does

The product has two clear surfaces. [The Amdahl pricing page](https://www.amdahl.ai/pricing) describes Sentiment Signals as an API that turns buyer conversations and CRM data into an enriched, indexed corpus that agents can query. Its Evals API grades GTM output against won, lost and stalled deals. Amdahl says MCP is included at the same usage rates, self-hosting is available per deployment, monthly spending caps are supported, and annual pricing can hold the rate.

## Why I’d look closer

The public pricing page is unusually concrete: [Sentiment Signals starts at $0.017 per run and Evals at $0.34 per run](https://www.amdahl.ai/pricing). That makes the first model easy to reason about. The harder question is what a “run” covers, how enrichment is maintained, and how the score relates to a real revenue decision. The company profile also carries a company-reported $4m pipeline claim and the homepage references a 12,480-signal corpus; I’m treating those as product claims, not verified traction.

The [public company LinkedIn page](https://www.linkedin.com/company/amdahl-ai) reinforces the product positioning: Amdahl describes itself as a context layer for GTM work, not a replacement for the underlying CRM or agent. That is the right distinction. The value depends on whether the company’s data is good enough to make an evaluation meaningful, and whether the output changes behavior rather than becoming another dashboard.

The founders are building from the problem. [The official profile](https://speedrun.a16z.com/companies/amdahl) lists Annette Sung, Robert Khoury and Arya Soltanieh, while the public company material describes experience across enterprise GTM, data and AI infrastructure.

## What I’d ask

Which connectors are live? How are permissions enforced? What counts as a passing eval, and how quickly can a team get from raw data to a trustworthy first answer?

## My editorial take

Amdahl is a good fit for companies already running GTM agents and frustrated by generic evaluations. It is early infrastructure, so the buyer needs enough data volume and a real quality loop. If the team only wants a chatbot over a CRM, this is probably more machinery than it needs.

## Quick facts

| Field | Sourced detail |
| --- | --- |
| Accelerator | a16z Speedrun, SR006 |
| Industry | Sales / GTM; AI Infra |
| Public website | [amdahl.ai](https://www.amdahl.ai/) |
| Pricing | [Sentiment Signals from $0.017/run; Evals from $0.34/run](https://www.amdahl.ai/pricing) |
| Product surfaces | Sentiment Signals API, Evals API and MCP |
| Buyer gate | Data connectors, permissions, eval design and corpus quality need verification |

## Sources checked

| Volatile field | Source | Checked |
| --- | --- | --- |
| Membership/founders | [Speedrun profile](https://speedrun.a16z.com/companies/amdahl) | 2026-09-19 |
| Pricing | [Amdahl pricing](https://www.amdahl.ai/pricing) | 2026-09-19 |
| Public company search | [Amdahl LinkedIn](https://www.linkedin.com/company/amdahl-ai) | 2026-09-19 |


## Author disclosure

I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.