Good News, Bad News—and Why We Built the Safe & Together Digital Coach and Insight Engine
By David Mandel, CEO and Founder, Safe & Together Institute
When I’m talking to child protection leaders about their workforce, I would often tell them I had good news and I had bad news.
The good news: their workforce were frequently very knowledgeable about the domestic abuse dynamics in the families they worked with. They knew which father in which household was the one the entire family was frightened of. They knew which mother was holding a family together under conditions nobody should have to parent in. The raw skill, the knowledge, the mission orientation—it was already embedded in much of their workforce.
The bad news came in two parts.
That raw potential was not being realised, and it was costing the families—and the agency. For example, the workers weren’t documenting the most important parts of what they knew. Hampered by outdated frameworks, approaches, and processes, they weren’t describing the impact of the perpetrator’s behaviour on the children. They weren’t fully capturing the survivor’s protective efforts—particularly of minoritised mothers. They weren’t analysing and recording the cultural and social forces increasing her entrapment. They weren’t connecting the dots about the relationship between domestic abuse and the other issues in the file—the substance use, the mental health, the housing.
And they carried a “failure-to-protect” mindset that pointed them at the survivor’s behaviour instead of the perpetrator’s, which made them less efficient, less effective, less ethical, and less safe in their practice.
Leaders Face the Same Split
My conversations with child protection leaders about their own work often run parallel.
The good news is that most leaders now understand that domestic abuse is core child abuse business—not a specialist referral or something that can be referred to a DV agency to “fix” with counselling for the survivor. Based on prevalence, lethality, and complexity, having a domestic abuse–informed system is central to achieving the benchmarks you are actually measured on—reducing out-of-home placements, building case plans that change something, and closing the disparity gap in removals from Indigenous, First Nations, and racialised families.
I often had to argue that 20 years ago when I started. I mostly don’t have to argue about it now. That is real progress.
The bad news—and none of this will be news to you—is this: You, as a leader, are saddled with antiquated policies and systems that were built for a different kind of case—physical abuse, sexual abuse, other forms of neglect. Your forms have a field for the maltreatment type and no field for the pattern. You are dealing with a siloed service delivery system, where the information that would make the case move forward toward safety and wellbeing sits in four agencies that don’t talk. You have a high staff turnover rate, which means you are re-teaching the hardest practice in child protection to a workforce that partially replaces itself while you’re doing it. And you are hiring from a social work education system that is still not effectively teaching the intersection of domestic abuse and child maltreatment. Your new graduates arrive without it. You are the one who has to install it.
How the Model—and the Institute—Answered That Need
The Safe & Together Model was a response to that—the real need of families, and of systems, for a methodology that consistently improves outcomes for families and increases the efficacy of child-centred systems. Saving them time. Saving them money. And the most mission-critical result of all: saving the lives of adult and child survivors by improving the quality of practice and the overall paradigm a system brings to domestic abuse cases involving children.
Over the last 20 years, built on the Model, we have scaled a training organisation, the Safe & Together Institute, that delivers training to roughly 10,000 professionals a year worldwide. That has a network of 100+ Partner Agencies and 600+ Certified Trainers moving as one to champion this approach in their own communities. That has a growing body of independent research behind it. That has moved well beyond child protection into family courts, substance use and mental health services, domestic abuse services, and child and family agencies. And that has demonstrated its effectiveness as a vehicle for practice change—a common framework and language that improves coordination inside an organisation and at the multi-agency table.
I am proud of that. I also want to say plainly what 20 years of it has taught me about its ceiling.
Training’s Ceiling
Training can teach a practitioner a new paradigm and provide an introduction to key domestic abuse–informed practice skills like interviewing, documentation, case planning, and assessment. But it can’t be expected to help the worker apply these new skills in the field, teach them how to apply them at specific moments in a specific case, or help them overcome blockers in confidence or skill development.
In our follow-up training data, across thousands of practitioners, we see strong indicators of practice improvement in partnering with survivors and in pattern-based approaches to perpetrators but not nearly enough consistent implementation of our Perpetrator Pattern Mapping Tool or strong enough increases in confidence and skills in intervening with perpetrators. And despite our strong investment in collecting data on training outcomes, we are unable, with our current tools, to capture information about consistent practice change over time in any area nor about outcomes for families and systems. Case file audits, independent evaluation of use of tools, and other key variables have been beyond our capacity to measure.
These gaps are not a failure of the training, our commitment to create new habits in workers, or our desire to know our impact. They’re a description of what training is. Four days gives a practitioner a framework, but it cannot reinforce new habits or coach the worker through the difficult moments, which is exactly what builds true skill and confidence. It cannot be sitting beside her while she writes the affidavit, or in the car before she knocks on the door, or in the 90 seconds before she opens her mouth in a case conference and has to choose between “the couple’s conflict” and “his pattern of coercive control.”
The Model lands or fails in those moments. For 20 years we’ve trained champions who are amazingly committed and effective in helping their peers, but we can’t train them fast enough, and there isn’t enough budget to have as many as are needed.
So after 20 years, I’m not ready to allow this ceiling to remain. I’m seeing now that the conditions are ripe to make the deep, sustained systems changes that everyone wants—from survivors, to their advocates, to child protection itself—at scale and within budget.
Why Is Now the Moment?
Technology is offering us new ways to deliver the Safe & Together Model framework at scale, in a cost-effective manner, embedded on the desktop of every worker who needs it. Always on. Always ready to provide consistent guidance and decision support.
And despite the fears, the challenges, and the real privacy constraints, AI is already in your workers’ practice, whether we know it or not and whether it’s helping or not. Your caseworkers are using AI in their hardest cases right now. They are pasting case material into a copilot or another untrained AI program at 2 a.m. and asking it what to do. Not because they are reckless, but because they are stuck, the supervisors are busy, and something is open in another tab that answers immediately.
What comes back is confidently, fluently wrong in a very particular direction. Generic AI does not know the Model. It has no reason to keep the perpetrator’s behaviour in frame. It drifts toward coaching the worker on how to manage the mother: safety plan her, assess her insight, evaluate her protective capacity. It pathologises trauma responses. It reproduces, in polished prose, precisely the failure pattern you want to avoid.
So the question in front of child protection leadership is not whether AI enters practice. It’s already in. The question is whether what enters is domestic abuse–informed, trained on your mandates and approaches, sanctioned, and visible to you—or generic, ungoverned, and invisible.
The Safe & Together Digital Coach Is Our Answer
The Safe & Together Digital Coach works the way any chat tool your staff already use works. A worker opens it and thinks out loud. She maps a perpetrator’s pattern, including how he is using the children as a tactic against their mother. She works a complex case—abuse plus substance use plus mental health plus poverty—without losing sight of the perpetrator’s pattern and its relationship to those other issues. She rehearses the hard conversation before it happens: the substance abuse counsellor, the kinship conversation, the family group conference, the yarn with an Aboriginal father. She uploads a de-identified note and asks what a Safe & Together lens would see in it.
That last one is modelled on how I used to teach domestic abuse consultants to apply the Model to their work. We’d sit down with a case file and I’d ask them what they saw—the perpetrator’s pattern, the mother’s protective efforts, what it meant for the case—before I told them what I saw, so we could compare notes. The Digital Coach asks first, too. It doesn’t hand down a verdict. It opens a conversation between what the worker saw and what the Model sees.
In alpha testing—from frontline workers up to service directors—the thing I found most promising was how many people used it to audit their own writing. In one exchange, a worker submitted documentation describing a mother as someone who “always picks bad men,” concluding that the father cared more about the children than she did. The Digital Coach named that documentation as destructive rather than domestic abuse–informed and held the line when the worker pushed back citing a decade of experience.
That is one exchange, not an outcome study. But consider what it means operationally. Practice drift that used to surface only in supervision—or far more often, not at all, until it was already in front of a magistrate—was caught before it reached the file.
None of this replaces your supervisors, your practice leads, or your Certified Trainers. It multiplies them. It is how a champion’s reach stops being limited by their calendar.
What the Evidence Does—and Doesn’t—Show
Now the part your finance person or your legislative body will ask about—and should.
Let me be precise about what the evidence does and does not show. There is no return-on-investment study on the Digital Coach itself. It’s too new—we are currently in the pilot phase, testing it under real-life conditions. What exists is a growing body of independent research on the practice change the Model has driven—the practice change the Digital Coach is built to deepen and sustain.
Here’s a portion of what we’ve found:
A study of the Safe & Together London Partnership, evaluated independently by London Metropolitan University’s Child and Woman Abuse Studies Unit, found three of the six boroughs recorded increased identification of domestic abuse; two recorded a decrease in child protection plans where domestic abuse was identified; and three recorded a decrease in looked-after children where it was identified.
The DV West children’s specialist program evaluation in Australia, where the Model was a core element, found that during the study period none of the statutory-involved families saw their children taken into care—in a cohort that was majority Aboriginal and in which many families had been assessed as high risk.
A University of Melbourne multi-agency triage pilot, using the Model to guide cross discipline conversations, directed almost 90% of child maltreatment referrals from law enforcement away from statutory child protection and toward community-led support.
Weigh your own budget worries against those three findings. One avoided out-of-home placement can fund a great deal of workforce capability in most jurisdictions I work in. Then add the costs that never appear in the out-of-home care line: the caseworker who leaves because she was pushed into practice she knew harmed a mother; the re-referral that cycles back through your front door because the case was misframed at intake; and the decision that was overturned by the courts because a child was removed without clear efforts to partner with the protective parent.
And then there’s the thing no training contract has ever been able to give you: the Insight Engine. It’s built to help you understand how practice is actually embedding across your organisation—audit and quality assurance, where skills are taking hold and where they’re not, practice strengths alongside the gaps, worker wellbeing alongside worker performance. Worker conversations stay private; what you see are patterns, not surveillance. If nobody in a region is using perpetrator-pattern mapping, that’s a practice gap you can target this quarter rather than discover in a child fatality report. When your minister or your oversight body asks what the training investment changed, you will be able to answer with practice-level evidence instead of attendance counts.
Now what the Digital Coach and Insight Engine is not.
It is not predictive analytics. Predictive scoring tells a worker how worried to be about a family. It does not teach her anything, it does not make her better next week, and when it’s wrong, it’s wrong about the people with the least power to contest it.
What I’m describing is the opposite of that. Real teaching. Real reminders about how to think critically about domestic abuse and children. Real-time access to tools and practice skills already proven to make a difference. Guidance that leaves the decision exactly where it belongs—with the caseworker, the supervisor, and the agency.
It also delivers two things I want to name directly: lived-experience leadership and data.
Lived Experience and Data, Built In
My co-founder, Ruth Reymundo Mandel, shares the creation and the governance of the Digital Coach. She brings the business vision and the discipline that turns our methodology into a digital platform an agency can actually procure, deploy, and be accountable for.
The Insight Engine—the measurement side of the tool—is one of her main areas of focus. Her question has been the one you would ask: How do we put real measures of practice change and case outcomes in front of the people who are held responsible for both? That’s the layer that closes the loop between the coaching a worker gets at 11 p.m. and the numbers you have to defend in front of your minister.
And she brings lived experience of domestic abuse, which means that design and functionality keep in mind both the survivor client—and the worker who is survivor herself.
The Ask
So here is the ask: Pull five of your own case files from the last quarter where domestic abuse was named. Read them the way I used to read them with consultants, and ask what you see. Is the perpetrator’s pattern documented as a pattern or as a series of incidents? Is the mother’s protective effort recorded as a strength or as a concern? Whose behaviour is the case plan actually about?
Then pull five of your substance use or mental health cases. Which ones missed domestic abuse indicators or failed to integrate knowing history of domestic abuse? How did the case plan with the mother fail to account for the perpetrator’s efforts to sabotage her treatment? When was substance use or mental health treatment substituted for men’s behaviour change?
Then find out what your workforce is currently doing with AI in those same cases. Not with a policy reminder. Actually ask them.
If what you find troubles you, register for a demo of the Digital Coach with us (if you’re in the Asia Pacific region, click here; if you’re in North America or Europe, click here), and we’ll put your own scenarios through it.
Then let’s talk about how we crash through that ceiling together.
Additional Resources
Podcast: Special Episode: A Safe & Together Expert at Your Side: The Digital Coach and Insight Engine
Safe & Together Institute’s domestic abuse–informed trainings
Safe & Together Institute’s upcoming events
David Mandel’s book Stop Blaming Mothers and Ignoring Fathers: How to Transform the Way We Keep Children Safe from Domestic Violence
