Win More Work Beyond the Buzz · Session 03 of 05
Golden Gate ALA
May 21, 2026
1-9 jump · ←→ navigate

Two to four people. A hundred attorneys.

Open · The Reframe
The Marketing Math
The team is structurally lean.
Every attorney wants pitch decks, thought leadership, RFP support, and a LinkedIn presence. The math has never worked. AI doesn't fix that by writing faster.
The Hidden Bottleneck
The bottleneck isn't writing. It's access.
Past RFPs. Past wins. Matter outcomes. Client intelligence. The asset base exists. It's just not reachable, not by humans on deadline, and not by AI either if the data isn't organized.
A Scene You Already Know · Mid-Size Litigation Boutique
A partner asks for "everything we've done" on healthcare antitrust in the last three years. RFP due in 18 hours.
Before

Three marketers searching email, SharePoint, individual hard drives, and old engagement letters. Five hours of triage. Two relevant matters surfaced. Both light on detail. The response goes out thin.

After

Marketing director types one question into the firm's internal AI: "Surface everything we've done on healthcare antitrust since 2023." Twelve matters returned in 90 seconds. Pitched. Cited. The response goes out strong.

From my client work, last 90 days. Firm and client names withheld.
The lift comes from access to what you already have. Not more content. Better retrieval.
The Shift
60 minutes. Five things. Origination vs. access · The RFP data problem · Content as a flywheel · Intelligence at scale · The lean marketing stack.

Origination or access.

Reframe · The Line That Matters
Origination · The Easy Answer
"AI writes the thing for us."
Blank-page generation. Draft a blog post. Write a pitch. Generate the RFP response. The work most marketers reach for first because it's what the demos show.
  • Bounded by what AI invents
  • Generic by default
  • Voice drift on every output
  • No firm IP advantage
Access · The Real Lift
"AI retrieves what we already have."
Multiply access to the firm's accumulated work product. Past RFPs. Matter outcomes. Win precedents. Client intelligence. Speeches. Alerts. The IP is already there. AI makes it reachable.
  • Bounded by what the firm has done
  • Specific by default
  • Firm voice preserved
  • Compounds with every new asset
Most teams ask AI to originate. The leverage is in access.
Four Workstreams Where This Plays Out
01
RFPs and Pitches
Past responses are the asset. Are they reachable?
02
Content
One asset becomes ten. Atomize, not originate.
03
Prospect Intelligence
Monitor what no human can monitor at scale.
04
Competitive Intelligence
Track lateral moves, GC changes, market signals.

The data problem before the AI problem.

RFPs · The Maturity Curve

AI can write a strong RFP response if it can read your past RFPs. Most firms cannot read their own past RFPs. That's the problem.

Level 0
Scattered
Past responses live in individual marketers' email, hard drives, and memory. No central index.
Most firms start here
Level 1
Centralized but Unsorted
A SharePoint folder exists. Files are named inconsistently. No tags. Nobody searches it twice.
Where most firms sit
Level 2
Tagged and Organized
Tagged by practice area, client industry, matter size, outcome. The work that has to happen before AI helps.
The foundational work
Level 3
AI-Readable
Indexed for retrieval. Queryable in natural language. "Show me everything on healthcare antitrust since 2023" returns answers in seconds.
Where the lift starts
Level 4
Agent-Accessible
An internal agent finds, retrieves, drafts a first cut, and cites sources. The marketer reviews and edits, doesn't search.
Where the compounding happens
Recent Engagement · Am Law 200 Firm
What changed once a 4-year RFP archive moved from Level 1 to Level 3.
5hrs
Average time to surface relevant precedent. Was 12 hours.
3x
Matters cited per response. From 4 to 12 on a typical pitch.
90days
From tagging kickoff to AI-readable archive. Most of it data work.
From my client work, last 90 days. Firm and client names withheld. Numbers anonymized to nearest round figure.

One asset. Ten outputs.

Atomize · Content as a Flywheel
The Source Asset
A 45-minute CLE webinar.
Two partners on stage. A topic the firm wants to own. One recording, one transcript, one set of slides. The raw material most firms record once and never touch again.
What Atomization Produces
↓ Atomized via AI ↓
01 LinkedIn post · practical takeaway
02 LinkedIn post · counter-intuitive insight
03 LinkedIn post · question hook
04 60-second highlight clip
05 Quote graphic for socials
06 Practice area client alert
07 Blog post recap
08 Attendee follow-up email
09 Newsletter section
10 Partner LinkedIn ghost-draft
A Working Example · Outside the Legal Vertical

WATT Poultry is an editorial publication. Their atomization runs through a codified set of voice rules the AI follows on every output.

Each article produces three LinkedIn posts at different angles. Industry insider tone. Value-first framing. Expert authority without sales speak. The voice rules are non-negotiable, written down, and enforced on every generation.

The point isn't poultry. The point is that a small content team built a system where AI atomizes consistently without voice drift. The same model works for a law firm. The voice rules are the asset. The atomization is the throughput.

The Risk

Without codified voice rules, atomization produces ten outputs that sound like AI wrote them. The lift evaporates.

What no human team can monitor at scale.

Intel · Prospect, Account, Competitive
Category 01

Prospect Intelligence

  • Monitor 50 target companies' 10-K and 8-K filings for litigation language
  • Surface M&A signals before they're public news
  • Detect regulatory enforcement actions in industries you cover
  • Track expansion into jurisdictions where you have presence
Category 02

Account Intelligence

  • Watch your top 30 clients for GC turnover and chief legal officer changes
  • Surface news about internal restructurings that change legal spend
  • Identify when a client lateral-hires senior in-house counsel from a competitor's industry
  • Flag client moves into adjacent markets where you have niche capability
Category 03

Competitive Intelligence

  • Track competitor lateral moves at the partner level
  • Monitor competitor practice group launches and rebrands
  • Detect when competitors win matters in your strongest verticals
  • Surface competitor thought leadership patterns and gaps
Campaigns Grounded in Data, Not Instinct
"Let's host a CLE on tariff impacts in healthcare M&A for our top 30 healthcare clients."
Instinct

The marketing team blasts the invite to everyone in the healthcare segment of the CRM. Three out of 30 register. Two are existing clients who'd come anyway.

Data

AI surfaces the 11 of 30 clients with material tariff exposure and active M&A pipeline. Eight register. Six are net-new conversations. One becomes a matter within 60 days.

Audit first. Then buy.

Stack · The Lean Marketing Setup
Marketing AI Features You're Likely Paying For

Before you add a vendor, run the audit.

  • Copilot in Microsoft 365 drafts, summaries, research
  • Salesforce or HubSpot AI lead scoring, account signals
  • InterAction with AI relationship intelligence
  • LinkedIn Sales Navigator account intent signals
  • iManage Insight+ matter-level search
  • Foundation or Passle experience management
  • JDsupra and Lexology distribution intelligence
  • Bloomberg Law, Lex Machina litigation analytics
  • Zoom AI Companion meeting summaries
Most firms have five of these and use one.
Three Cost Tiers · For What's Worth Adding
Tier 1 · Start Here
Low five figures annually
Enterprise Claude or ChatGPT seats for the marketing team. A structured prompt library. A codified voice guide. Mostly process work on top of what the firm already has.
Covers atomization, drafting, prospect research
Tier 2 · Once Tier 1 Works
Mid five figures annually
Add a specialized marketing AI vendor. Foundation-tier experience management with AI surfacing. Competitive intelligence subscriptions with structured monitoring.
Covers intel monitoring, experience search
Tier 3 · Capital Decision
Six figures, year one
An internal AI agent built on the firm's RFP archive, matter outcomes, and content corpus. Requires the data work first. Six to nine months. Pays back across every pitch and proposal afterward.
Covers the Level 3 to Level 4 RFP jump

What goes wrong. Before it does.

Reality · Five Failure Modes

Every failure mode below has happened in a real firm in the last year. Name them out loud before the rollout, and most are preventable.

1
Hallucinated win records in an RFP
AI confidently cites a matter the firm didn't actually handle, or attributes a win to the wrong lead partner. The most damaging failure mode because it surfaces to a client.
Mitigation

Restrict AI retrieval to verified, tagged source documents only. Block open-web augmentation on RFP drafts. Human signoff on every cited matter.

2
Brand voice drift across atomized content
Ten LinkedIn posts go out. Each one sounds slightly more like ChatGPT than the last. The firm's voice erodes one output at a time, and nobody notices until a partner does.
Mitigation

Codify voice rules in writing. Run every atomization through a fixed voice prompt. Quarterly audit of published output against the firm's actual voice.

3
Confidentiality exposure on shared AI tools
A marketer pastes a confidential RFP into a public AI tool to get a draft response. The data is now in someone else's training corpus. The firm finds out months later.
Mitigation

Enterprise AI seats only, with data not retained. Block consumer AI URLs at the firm level. Train marketers on what's allowed and what isn't.

4
Attribution errors in summaries
AI summarizes a matter and credits the wrong lead partner, the wrong associate, or the wrong client industry. Internal credibility erodes faster than external.
Mitigation

Use structured matter data, not free text, as the source for attribution. If you can't tag it, AI can't safely cite it.

5
RFPs that read like AI wrote them
The most subtle and the most common. AI drafts the response in generic professional prose. The marketer ships it without layering in firm voice or partner perspective. The client knows immediately.
Mitigation

Treat the AI draft as the first 50 percent, not the final 90. Mandatory human editing pass on every external-facing RFP, with named editorial owners.

Your turn.

Live Sentiment Survey
Live Sentiment Survey

Where is your team on this?

Your answers shape Sessions 4 and 5.
Launch Poll
What We're Asking
Question 1
Where do your past RFP responses currently live?
Question 2
What share of marketing time goes to RFPs and pitches?
Question 3
Have you codified a firm voice guide for AI-generated content?
Question 4
Which workstream would AI help your team most with right now?
Question 5
One word: what's blocking your marketing team most today?
After The Poll
Pull two or three sharp threads live. Full readout goes to registrants.

What's next.

Series Roadmap + One Action
Two Sessions Remain
5 RFPs
One Thing to Do This Week
Pull your last five RFP responses.
Then ask three questions.
"Can AI read these? Can it find them by client industry? Can it find them by outcome?"
If the answer to any one is no, that's your first 90 days. Tagging, organizing, and indexing the archive is the work that makes everything downstream possible. Nothing else matters until that's done.
Thank you. Questions? Drop them in the chat. Spencer X. Smith · AmpliphI · spencerXsmith.com