# Amanda DiCicco Stilman

Amanda DiCicco Stilman works on enterprise AI products and content focused on governance, compliance, and practical deployment in regulated industries. Their posts discuss AISquared products and frameworks for AI controls, data governance, and automating proposal work.

- Profile: https://b2binfluencer.marketing/creators/amanda-dicicco-stilman/
- LinkedIn: https://www.linkedin.com/in/amanda-dicicco-stilman
- LinkedIn followers: 1,630
- Topics: artificial-intelligence, ai-governance, data-governance
- Based on 13 dated posts, latest 2026-10-02

## Summary

They are most useful for enterprise AI governance, compliance, and deployment content for regulated-industry buyers.

- Best for: A regulated enterprise software brand wanting governance and compliance-ready AI deployment content for security, GRC, procurement buyers.
- Not for: Short-form video reach or product demos for a large practitioner audience.
- Check first: Confirm audience demographics and recent results on comparable governance and compliance content.

### Takeaways

- A seven-layer AI controls framework names evidence to request from vendors.
- Weak data lineage can make an accurate AI answer indefensible.
- Bolt Instruct routes routine work to smaller models and adds guardrails.

## Main ideas

- **Governance is what you can prove, not what you were promised**: The most detailed post argues that buyers — examiners, auditors, underwriters — will ask what an organisation can demonstrate about its AI, not what a vendor assured it. The response is a seven-layer AI controls framework covering systems of record; connectivity and access control; data processing and context; workflow orchestration; policy and governance; delivery and embedding; and observability and continuous improvement. For each layer, the guide sets out questions to put to a vendor, documentation to request, and answers that should make a buyer walk away. The post argues that a governance platform proves its value by producing artefacts such as scoped credentials, tamper-evident logs, versioned policies and drift dashboards. It presents the framework as vendor-neutral: something to apply to any vendor, including the author's own company.
- **When AI gives a bad answer, look at the data layer, not the model**: A recurring argument: when a model produces a wrong output, the instinct is to blame the model, but the real problem usually sits one layer down, in the systems of record an enterprise has already decided to trust — ERP, CRM, HRIS, case management and data warehouses. The post describes these as years of accumulated organisational truth that the business already runs on. In regulated settings, the standard shifts from accuracy to provenance: an auditor wants to know where an answer came from. If the AI draws on an ungoverned pipeline, the answer cannot be defended, so the deployment fails on defensibility rather than performance. The stated remedy has three parts: documented lineage from output back to source, governed access designed in from the start, and a clear hierarchy for when systems disagree.
- **Open models as a procurement requirement, not a technical preference**: In a product-architecture post aimed at regulated industries, the argument is that enterprises cannot deploy AI there without first clearing procurement, legal and compliance review, and that model openness is part of passing that review. The product described, Bolt Instruct, is positioned as an orchestration layer that routes requests to the right model, enforces guardrails on inputs and outputs, detects PII and jailbreak attempts, and produces structured outputs for high-volume work such as retrieval-augmented generation. Routine tasks go to smaller purpose-built models, with frontier models reserved for work that needs them. Because Ai2 publishes Olmo's training data, code and architecture, the post argues that customers can trace what they are building on through compliance review.
- **Applying the same logic to unglamorous operational work: bids and proposals**: A shorter, more conversational post applies the same problem-first approach to revenue and bid teams rather than security leaders. It describes responding to RFPs and ITBs as time-consuming, fragmented and reactive: teams miss relevant bids, chase low-probability opportunities and spend days drafting proposals from scratch. The post presents Sparx Proposal Intelligence as covering bid discovery, win-probability scoring and proposal generation, so teams can focus on opportunities most likely to convert. It closes by asking readers how they handle bids today.
- **A consistent method — and a tension between neutral guidance and product promotion**: Across the posts, the structure repeats: a short problem statement, then a framework, list or architecture carrying the substance, followed by a link to a guide or product page and often a question to the reader. The vocabulary stays rooted in enterprise buyers, compliance review and procurement. One post offers a framework to use on vendors including the author's own company, while others describe that company's products with concrete performance, cost and efficiency claims.

## Work to read

- [The CISO's AI Governance Guide: a seven-layer AI controls framework](https://www.linkedin.com/posts/amanda-dicicco-stilman_aigovernance-ciso-aisecurity-activity-7511804152117886976-7R-9)
- [Bolt Instruct inside UNIFI: orchestration, guardrails, and open models in regulated industries](https://www.linkedin.com/posts/amanda-dicicco-stilman_bilt-instruct-unifi-activity-7498410213226598400-GYD1)
- [Systems of record: why AI deployments fail on defensibility, not performance](https://www.linkedin.com/posts/amanda-dicicco-stilman_enterpriseai-ai-aisquared-activity-7507888903786356737-R1qh)
- [Sparx Proposal Intelligence: automating bid discovery, scoring, and drafting](https://www.linkedin.com/posts/amanda-dicicco-stilman_procurement-rfp-aiautomation-activity-7450593106015522816-e_fZ)

## Brand relationships

- **AISquared** (employment)
- **Ai2** (mention)

## Partnerships

The posts do not say that Amanda DiCicco Stilman offers creator partnerships or sponsorships. All five promote employer-side enterprise AI products and frameworks: Bolt Instruct inside the UNIFI platform, the seven-layer AI Controls Framework and The CISO's AI Governance Guide, and Sparx Proposal Intelligence. They include no sponsored-content disclosure, rate card, deliverable or format specification, audience or performance figures, or terms of engagement. Brands planning an AI-governance and compliance-ready deployment campaign should treat this as an employee-advocacy and thought-leadership channel, not a standing paid-creator inventory, and expect commercial conversations to go through the employer organization rather than an independent creator deal. Ask the employer whether it offers an official partnership route; the posts provide no inquiry link, and the general homepage is not a substitute.

## Alternatives

- Akhilesh Perla: Akhilesh Perla is the stronger choice for a campaign about the data-governance layer underneath AI deployment, rather than a framework layered on top. Both posts argue that the problem is the data, not the model: fragmented systems, untrusted or poorly governed data, and AI that works in demos but collapses in real workflows. The work gets concrete about the stack—MuleSoft for integration, Informatica for data quality and governance, Agentforce for execution—and what buyers get from it: trusted data by default, metadata and lineage across the enterprise, and agents that start workflows and enforce policy as they act. A sponsor commissioning content on lineage, master data, and controlled access inside its own platform will find more architectural detail here than in a vendor-neutral controls checklist. Published 19 November 2025 and 23 June 2026.
- Jacob Riff: Jacob Riff is the better pick for a campaign about compliance readiness for a small team trying to get through an audit, rather than a large regulated enterprise buying an AI platform. The two posts do what the Amanda DiCicco Stilman corpus does not: walk through a process end to end. One covers the SOC 2 path in working detail—around twenty policies, roughly 100 to 150 controls, vendor due diligence, risk assessment, penetration testing, and the difference between a Type I report covering a single day and a Type II run over a three-to-six-month observation window. The other meets the strongest objection head-on—that a compliance platform is a waste of money—and concedes that a spreadsheet and a shared drive can get an attestation. Published 20 August 2026 and 6 August 2026.
- Gaurav Bhattacharya: Gaurav Bhattacharya fits a campaign focused on deploying AI inside production workflows rather than governing it for an audit. The stronger post names the hard problems directly—orchestration, reliability, memory, permissions, governance, recovery systems, observability, and trust—and argues that enterprise workflows are limited by orchestration, not intelligence, before sketching a signals-to-reasoning-to-action-to-human-oversight stack. The other makes a distribution argument about AI commoditizing execution. For a sponsor seeking content on what actually breaks when agents run in production, that operational framing sits closer to engineering reality than a controls framework. Published 18 May 2026 and 22 May 2026.

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