The Anatomy of Success: Key Insights from AI-Driven In-House Teams

What are the best practices for deploying AI? What are the key challenges in AI deployment and how to avoid them? How to pick a good AI use case? How to successfully leverage GenAI in organisations?

The Anatomy of Success: Key Insights from AI-Driven In-House Teams
Photo by Nik on Unsplash

I spent the last four weeks looking into the AI implementation strategy deployed by 20 different organisations across a range of industries and geographies. These companies, all navigating the B2B landscape, shared a commitment to enabling fast business growth and innovation. I observed some key trends and telling indicators of success (or failure).

I’m sharing these insights to spotlight the strategies employed by these teams so that they may also serve as a roadmap for other organisations looking to harness the transformative power of AI.

What I’ve learnt:

1. Clear Leadership and Responsibility Move the Needle

When senior leaders - often with direct access to the CEO - spearhead these initiatives, impact is visibly felt, and quickly too. Having decision-makers who are not only invested in the technology but who also possess the influence to drive its adoption is key to moving the needle in-house.

A case in point is a sales intelligence B2B company based in Israel and their fast and effective deployment of AI for sales enablement. The close involvement of a Senior Legal Counsel and General Counsel illustrated the transformative power of having leaders who are not only visionary but also strategically positioned close to the helm.

A critical success factor for the impactful deployment of AI and tech more generally, is the presence of clear leadership and responsibility.

In contrast, adoption can be more of an uphill battle where projects are led by individuals with limited organisational influence. Where individuals have limited authority, this can be a major challenge to effecting meaningful change. Common gaps here are a lack of strategic direction and executive support.

2. Agile Teams Can Harness Chaos as a Catalyst for AI Acceleration

In one particular case study, I observed an incredible display of agility, resilience and visionary innovation in the midst of national turmoil and conflict. Virtually overnight, 33% of a Legal team was conscripted in dire circumstances. What followed next was nothing short of spectacular in terms of creative thinking and human resilience by fast tracking the deployment of AI bots to absorb workload previously fulfilled by a full team. This underscores the agility and adaptability of successful teams.

Despite significant disruptions, this team recognised the unique opportunity to leverage AI to mitigate the impact of their stretched resources. This ability to pivot and adapt to external challenges is a hallmark of resilient and forward-thinking teams.

The most effective teams are those that problem-solve on the go, quickly identifying opportunities for significant gains and adopting the right strategies to seize them. This dynamic environment requires a blend of rapid action and patience.

The initial phases of implementing solutions can feel cumbersome and time-consuming, but the ability to anticipate the long-term benefits and stay the course is testament to a team's strategic foresight and commitment to innovation. In navigating potential challenges, the patience for longer-term wins becomes as crucial as the agility to adapt to immediate circumstances, underscoring the depth of strategic thinking that underpins successful AI integration. I will also say that with the right solution, a good starting point and the right expert guidance, adoption can be a breeze!

3. Simple, Effective Solutions Are Excellent Starting Points

There’s incredible power in starting small. The most impactful in-house teams tend to have an initial focus on simple, yet effective solutions. For example, 73% of organisations I looked at have set up a first bot using a Frequently Asked Questions (FAQs) document. This applies across various expert fields including Privacy, Compliance, InfoSec, Legal.

The approach to getting a FAQ document in place tends to vary across organisations too - for some, this type of document is a company staple and for others, a FAQ document is put together specifically for these bots, as and when use cases are identified. There is no right or wrong answer here but the common denominator is the fact that all that is needed for effective automation is consolidated context to form the basis of the bot’s output.

Many of our customers choose to leverage pre-existing internal documents to inform their context creation. Leveraging existing resources is definitely a fast-track to creating immediate value for the business.

4. Incremental Improvement and User Engagement Must be Tracked

Successful teams recognise that an ongoing commitment to refining AI solutions based on user feedback and real-world application is a hallmark of effective deployment.

This process of iterative improvement ensures that AI tools evolve in response to user needs and operational insights, thereby maximising their relevance and impact. In practice, this means paying extra diligent attention to the experiences of early adopters. The feedback loops generated there are invaluable for tailoring AI tools to deliver truly practical and scalable solutions.

5. Learning Curves are Okay and Expected

Adopting and integrating AI successfully comes with recognising the learning curve that accompanies this motion. I’ve been intrigued by the range of reactions from various types of organisations and teams. Behavioural change has manifested in different ways. I identified three different buckets:

  1. impact is immediate whilst the bots go nearly unnoticed where AI is deployed in existing high traffic areas and, more specifically, where conversations are already happening between expert teams and the business. This scenario involves the least disruption in terms of behavioural adaptation because tasks are being automated seamlessly in an existing high traffic area. There’s not much that’s changing from the business’ perspective, except, of course, the increased speed at which it’s receiving support;
  2. adoption can be very slow where integrating a new technology requires a complete overhaul of existing processes and communication channels. This is often the case with all-in-one platform solutions which (context dependant) can be akin to replacing a whole kitchen when only the sink drain was blocked;
  3. human insecurities throw spanners in the works. Building trust takes longer than anticipated because i) experts are reluctant to ‘let go’ of elements of the job that have been part of their working identity for so long and b) the business is hesitant to confidently self-serve from the get go because of reliability/accuracy issues or simply because this is new and there is an element of risk involved in charting new territories.

Whichever bucket you fall into, acknowledging and addressing these learning curves, through user education and feedback loops, is critical for ensuring the tool’s effectiveness and user satisfaction.

6. If One Team Does It Right, Others Will Follow

I found that if one team pulls off AI adoption effectively by proving success with a use case that has clear benefits for the organisation as a whole (for example, removing bottlenecks for sales teams and therefore speeding up the sales cycle by 5x), other departments are quick to feel inspired and catch on to the momentum.

Across most organisations I looked at, it is evident that cross-departmental collaboration is a key determinant of success or failure.

The reverse is also true - a poorly chosen use case (perhaps one that’s too complex or simply not suitable for automation yet, such as queries that require deep expert knowledge and judgment on case-by-case basis) that does not yield positive results can kill any momentum and feed a sense of techno-phobia in an organisation.

7. It All Starts (or Breaks) with the First Use Case

This leads me on to my next point which is selecting the right use cases for AI deployment. This is really important. Success often hinges on identifying high volume and high-traffic workflows where AI can have an immediate and measurable impact. How do you know if you have a good use case? If you answer yes to two or more questions in this list below, chances are that you do:

  • Is your organisation often dealing with a high volume of requests that tend to come up again and again?
  • Are these requests usually fulfilled by expert teams?
  • Are these queries generally low value (for example, admin, policy based queries, RFPs, security questionnaires, DPA reviews, NDA reviews, procurement contract reviews or otherwise non-complex tasks)?
  • If these queries are slow to be answered, does this cause delays for business workflows?

8. External Challenges Don’t Mean It’s the End of the Road

Some of the organisations I investigated faced their share of both external challenges and internal challenges that adversely impacted on their AI deployment. This includes national conflicts, rounds of layoffs, market instability, restructuring, multiple changes in key leadership etc. The underlying frame for successful teams, however, was the perseverance and flexibility displayed in navigating constraints whether they were external pressures, organisational bureaucracy, or technical constraints.

Having the discipline to carve out attention to problem-solving, when the urge to solely focus on fire fighting can be so strong, is one that really pays off in securing efficiency gains for the entire organisation in the medium to long term.

All in all...

All things considered, it's evident that the path to AI integration is multifaceted. Strategic foresight, leadership, and a culture of innovation and resilience are as (if not way more) important as a bit of technological flair. It's through the confluence of clear vision, cross-departmental synergy, and an unwavering commitment to continuous improvement that in-house teams can truly harness the transformative potential of AI.