In-House Legal Department Outlook: Lessons From GC and CLO Interviews

Legal departments are being asked to make several connected decisions: what work stays in-house, when outside counsel is worth the cost, how to develop lawyers, and where AI can take on repeatable work without replacing legal judgment.

GC AI is an enterprise legal AI platform for in-house counsel, trusted by 2,200+ legal teams. On our CZ and Friends podcast, GC AI CEO Cecilia Ziniti, a three-time general counsel, interviews current and former GCs and CLOs about how they are making those choices in practice. Five of those conversations offer an in-house legal department outlook grounded in management choices.

Taken together, the five interviews point to a practical principle: define the service the business needs before deciding how to staff it, outsource it, or apply AI. They are illustrative rather than a representative survey of legal departments.

Key Takeaways

• Start with the service the business needs and repair the handoff before adding a process or tool.
• Give outside counsel a defined role based on specialist expertise, capacity, and escalation triggers.
• Delegate with checkpoints so lawyers build judgment with support.
• Plan staffing and AI rollout together, with time for review and training.
• Protect regular contact with business teams as routine work becomes self-service.
• Measure one repeatable workflow before expanding it.

Start With the Business Problem Legal Needs to Solve

Bjarne Tellmann, former General Counsel at Pearson and Haleon, described the legal function as a service business with one client:

I kind of imagined myself as a CEO of this legal services business that has one client, Pearson.

When Tellmann arrived at Pearson in 2014, he found a legal department whose structure no longer fit the company. Before a GC adds a legal process or AI workflow, define the service the business needs and who should deliver it.

At a smaller company, Rachel Harris, General Counsel and AI Governance and Privacy Officer at Suzy, a consumer insights platform, found that part of the bottleneck existed before legal ever reviewed the contract. Account executives were sending services contracts when they were selling a SaaS product, so the problem was not simply legal's review speed; the sales-to-legal handoff needed work.

Harris framed the problem this way:

The real question becomes, have we not integrated ourselves into this workflow appropriately, to where once it gets to us, it feels like a bottleneck?

She also described putting contract guidance into Salesforce when a dedicated contract lifecycle management platform was outside her budget. Sellers could use a central reference in the system where they worked. She scoped that approach to a startup of about 100 people and acknowledged that a larger organization might need a different setup.

Tellmann was reconsidering a global department's structure; Harris was repairing a sales-to-legal handoff. In both cases, the useful starting point was the work itself. Trace a recent request back to the person who initiated it: what did they know, and what information did legal have to chase down?

If a request reaches legal without the information the lawyer needs, fix the handoff before trying to make review faster.

Give Outside Counsel a Defined Job

Chuck Kable, General Counsel and Corporate Secretary at Innovative Renal Care, showed why cutting outside counsel spend is not the same as eliminating outside counsel. At his previous company, Emerus, he eliminated outside spend on work the team had determined it could handle internally. Sophisticated joint venture work still involved external firms.

The distinction between in-house counsel and a law firm helps explain that choice. In-house counsel works inside the business and advises with its ongoing priorities in view. External lawyers contribute specialist expertise or additional capacity when a matter requires it. At Pearson, Tellmann matched outside providers to the work and reserved specialist firms for matters that needed their expertise.

For a department plan, decide which work the in-house team can own and which work genuinely needs outside expertise or capacity. Lower outside-counsel spend can reflect a sensible shift of work in-house, but the GC still needs to know who can take that work and what support they require.

For each recurring category, answer three questions:

• Who owns the work internally, and what expertise can the team supply?
• Why would an external lawyer be needed: specialist knowledge or additional capacity?
• What event or issue triggers escalation?

Those decisions give outside counsel management a concrete basis. They also give your next technology evaluation a specific question: can the team complete the proposed internal work to the standard you need?

Delegate With Checkpoints and Support

Anirma Gupta, former Unity Chief Legal Officer, described assigning a cybersecurity reporting project to a lawyer two levels below her. She helped define the result and identify the people the lawyer needed to involve.

The assignment came with continued support. Gupta used checkpoints to review progress and ask whether the lawyer needed help with roadblocks or executive pushback. Her questions included:

Are they hitting any roadblocks? Do they need any further support?

Gupta's example turns delegation into three concrete questions: what can the lawyer decide, when should they check in, and what should come back immediately? Put those answers in the project brief, and keep enough time available to discuss the lawyer's reasoning and step in when a project crosses departments.

Gupta also described giving developing leaders exposure to management and the board, with less support as they gained experience. When assigning stretch work, treat that development as one of the outcomes: greater decision-making responsibility, broader exposure, and less support over time.

Plan Legal Department Staffing and AI Rollout Together

Gupta predicted smaller legal departments and law firms as AI develops.

Ron Bell, Chief Legal and Administrative Officer at Collective Health, made a different, more immediate point: legal still needs enough capacity to serve the business. A commercial legal team unable to keep pace with deal flow can slow the business.

These ideas answer different questions. Gupta offers a forecast about how AI may change staffing over time; Bell argues that today's team still has to be staffed to the work the business needs. A GC can test whether AI changes the workload, but staffing decisions still need evidence about demand, service levels, and capacity.

Bell also raised a training concern. Lawyers starting with AI may need deliberate mentoring to develop the instincts that help them question its answers. A 2025 Nature Reviews Psychology review notes that large language models can support decision-making while still producing biases and hallucinations that constrain real-world use.

Gupta put the human role plainly:

But the real value add for us lawyers is at the judgment layer.

When junior lawyers use AI, make the reasoning part of the assignment: ask them to explain the recommendation, what they verified, and how they reached it.

Rolling out AI requires time and staffing of its own. Kable said he had postponed a project aimed at reducing outside counsel spend while the team absorbed other changes. He also envisioned AI helping support staff prepare drafts for attorney discussion and review:

The AI tools like GC AI are a force multiplier for my support level folks, for my paralegal team.

His longer-term goal was to hand off certain work to that support level after the pilot. At the time of Kable's interview, that support-staff pilot was still being developed.

Separate what the team is doing now from what it may expand later:

• Assign someone to maintain the review standards.
• Reserve attorney time for checking early results and coaching the people using them.
• Set a review date and decide what evidence would justify expansion, revision, or stopping.

Budget for the implementation and review time the team will need.

Preserve the Conversations That Help Legal Advise the Business

Kable also warned that self-service can remove some of the routine requests that create informal contact with business leaders. If those requests disappear, legal may need to create that contact deliberately.

Harris's earlier sales-handoff example shows why the contact matters. Understanding how another team works can reveal a problem before it becomes a legal request.

If you introduce self-service contracts, set a recurring conversation with the sales or product team using them. Ask what changed since the team designed the workflow, which exception they could not resolve, and which agreement they wish they had brought to legal earlier.

Track that feedback with the workflow's results. A process can handle familiar requests well while the business develops a new need.

Test One Contract Review Workflow With GC AI

The business and staffing questions above lead to a practical product question: where can legal apply its standards repeatedly while keeping a lawyer in the loop?

One way to test that with GC AI is to combine Playbooks with GC AI for Word. Playbooks apply a team's review positions to contracts, while GC AI for Word lets lawyers run those reviews and work with suggested redlines inside Microsoft Word. Counsel still owns the decision. The workflow makes the first pass repeatable so the lawyer can focus on deviations, context, and judgment.

You can test that approach with a recurring nondisclosure agreement (NDA) workflow. Choose an agreement type your team reviews often. The steps below give the reviewer a defined assignment and the supervising lawyer clear checkpoints.

1. Choose the work and the reviewers: Select an NDA category with established company positions. Assign a colleague to prepare the review and a lawyer to own the pilot. Select previously reviewed agreements against which to check the output.
2. Build from approved materials: Upload your current template or existing playbook in GC AI chat, along with relevant negotiation guidance. Explain your priorities, non-negotiables, and fallback positions. Our Playbook Builder Agent creates a structured draft and surfaces questions that require your judgment. Review the checks and resolve those questions before using it. The playbook creation guide explains the available inputs and editing process.
3. Run the review with deal context: Open an agreement in Word, select the playbook, and identify whether the agreement is on your form or the counterparty's. Add the business context needed for that review. Our Playbooks return analysis and proposed redlines, then label each check Pass, Fallback, or Flag.
4. Have the colleague explain the recommendation: Ask them to compare the output with the agreement and the approved positions. They should identify unresolved questions and explain each proposed escalation. The lawyer reviews that reasoning and checks the underlying language, including every provision our Playbooks label Pass.
5. Make and save the decision: The lawyer decides which changes to apply or revise in Word and follows the team's existing approval process before the agreement goes out. Record exceptions with the rationale for accepting or rejecting them.
6. Improve the next review: Correct unclear standards and test again. If a provision falls outside the playbook, Smart Checks can surface it; save useful checks to the playbook for future reviews. The guide to running a playbook explains the results and redline controls.

This GC AI Playbooks demo shows you how to run a playbook-based contract review:

For the pilot, record attorney review time and the corrections required. Ask the colleague which decisions they can now explain and where they still need help. Compare the results with your existing process before expanding the handoff.

Build the Department Around the Work

Taken together, these interviews point to a practical way to shape an in-house legal department: start with the work the business needs, decide who should own it, give lawyers room to develop judgment, and use AI where the work is repeatable enough to test and review.

A strong legal function can take on the right work without losing business context or lawyer judgment. Smaller teams, lower outside counsel spend, or more automation may follow, but those outcomes should follow from the work rather than drive the plan. Start with one narrow, repeatable workflow, measure what changes, and expand only when the results support it.